CVj = 2|f1j-f2j|/(f1j+f2j)

2.4.3.1 Data filtering using multiple gene data filters

Figure 2.4.3.1 Filtering using multiple scrollers. This example is of Cy3/Cy5 time series data. It filters normalized spot intensity of the Cy3 and Cy5 channels independently ([SI1:SI2] inside range) where low intensity spots are eliminated. It then filters out genes outside of the [R1:R2] ratio range.

Figure 2.4.3.2 Using the Positive Intensity data Filter. This allows removing negative data if the data contains negative intensity values (e.g. Some Affymetrix data has negative Average Difference values which could be read as Intensity for MAExplorer).

2.4.4 Plot menu

• Show Microarray - display the pseudoarray image for the current HP sample
• Scatter plots - display various scatter plots of selected pairs of HP samples
• Histograms - display both ratio and intensity histograms of HP sample data
• Expression profile plots - display expression profile plots of genes or gene subsets

You may switch between different representations of the microarray spot pseudoarray image. It may be viewed as several different types of pseudo images including an intensity gray value and a pseudo-color Red/Black/Green image for ratio (HP-X/HP-Y) and Zscore (HP-X - HP-Y) data. The p-Value results of comparing a HP-X 'set' with a HP-Y 'set' of samples, or the CV of the HP-EP 'list can be displayed as a color spectrum pseudoarray image.

Depending on the origin of the array data, it may have the same verisimilitude as the original arrays. Otherwise, it is displayed in a generic pseudoarray image containing grids that will fit the window - these are not the same as the original array image (see . However, the pseudoarrays are useful to getting a rough idea of the global changes in the data between arrays and how may genes pass the data filter.

When enabled using one of the commands in the Section 2.4.5 Clustering menu, cluster data appears as blue circles or squares drawn as overlays on the pseudoarray image. These options are discussed in the section on clustering. If you are doing clustering K-means clustering, the current cluster is displayed in the scatter plot if the latter is active.

Scatter plots, ratio and intensity histograms of the mean (HP-X/HP-Y) or (HP-X/HP-Y) 'set' data, or the F1/F2 or Cy3/Cy5 data. F1/F2 or Cy3/Cy5 plots are available if the data exists in your particular database. That might be the case with replicate spots or with Cy3/Cy5 data. If the normalization is set to a Zscore or log mean mode, it will compute Zscore scatter plots and histograms.

Clicking on spots in an array image or points in scatter plots sets the current gene and will bring up data on the gene or (optionally) access corresponding data from GenBank, UniGene, mAdb Clone, etc. databases in a popup Web browser. Clicking on a bin in a ratio or intensity histogram plot filters out all genes except for those in the range of that bin.

Expression profiles plots of selected genes or subsets of genes for all samples in the HP-E list. These are active plots with data reported when the user clicks in the plot.

Clicking on a spot (i.e. gene) in the microarray pseudo image or on a point (i.e. gene) in the scatter plot, it will define that gene as the "current gene" that is used in other operations. The current gene is indicated in both plots with a green circle around it. Similarly, you may modify the 'Edited Gene List' from either the pseudoarray image or the scatter plot. When viewing is enabled, it overlays those genes with magenta squares.

Figure 2.4.4 Plot menu - selecting Ratio Pseudoarray image. This displays a pseudocolor show in the scale on the left that indicates the ratio of the value of the HP-X sample / HP-Y sample (or 'sets' if the option to use HP-X and HP-Y 'sets' is enabled.) If The data is Cy3/Cy5 data, then this displays the ratio of the ratios using the current normalization. Various other pseudoarray image representations could be used.

2.4.4.1 Show microarray pseudoarray images menu

If the database that was loaded contains only one sample, the pseudoarray image display defaults to the pseudograyscale spot intensity mode. If there is at least one HP-X and one HP-Y sample, then the Pseudocolor HP-X/Y ratio or Zdiff mode is the initial default display. If there are duplicate spots for each gene, you may generate a Pseudocolor F1F2 ratio or Zdiff mode image. If you are using Cy3/Cy5 ratio data and the data is available as independent channels for each HP, then you may plot Cy3 vs Cy5 for individual samples.

When available on the database server, the original image may be displayed in a separate popup Web browser.

• Pseudograyscale intensity [RB] - display a pseudograyscale microarray pseudoarray image for the current H.P. where higher intensity spots are mapped to black.
• Pseudocolor Red(X)-Yellow-Green(Y) HP-XY ratio or Zdiff [RB] - display the HP-X and HP-Y microarrays as a pseudocolor image as either a ratio (HP-X/HP-Y) or Zdiff (HP-X - HP-Y) if Zscore normalization is in effect. The HP-X (red) and HP-Y (green) components are added together so that high total intensity values are yellow and low values are black. High HP-X/HP-Y ratio values are more red and low values are more green.
• Pseudocolor Red(Cy5)-Yellow-Green(Cy3) Cy3/Cy5 data [RB] - display the Cy3 and Cy5 microarray channels as a pseudocolor array image. The the Cy5 (red) and Cy3 (green) components are added together so that high total intensity values are yellow and low values are black. High Cy3/Cy5 ratio values are more green and low values are more red. Note: this is only available with Cy3, Cy5 ratio data. Note: if the "Use ratio median corrections" is enabled, the Cy3 channel is scaled using the to the domain of the Cy5 channel by (median Cy5/median Cy3).
• Pseudocolor HP-X/Y ratio or Zdiff [RB] - display the HP-X and HP-Y data as a pseudocolor array image as either a ratio (HP-X/HP-Y) or Zdiff (HP-X - HP-Y) if Zscore normalization is in effect. The high ratios or Zscores are red and low values are green. Values in the middle are black.
• Pseudocolor (HP-X,HP-Y) 'sets' p-value [RB] - displays each spot as a pseudocolor proportional to the p-Value in a t-Test of the HP-X 'set' vs HP-Y 'set'. This can only be used when the "Use HP 'sets' ..." option is enabled and there are at least 2 samples in both the HP-X 'set' and the HP-Y 'set'. Note that unless proper normalization and filtering is used to remove poor quality data, some of the p-Values will not be significant and there will be a high false-positive rate. Use this display with that in mind.
• Pseudocolor HP-EP 'list' CV (Coefficient of Variation) [RB] - displays each spot as a pseudocolor proportional to the CV of the spots within the subset of samples in the current HP-EP 'list'. It is useful to look at the variation in a set of replicate samples. Use the (Samples | Choose HP-X, HP-Y and HP-E samples) command to define the HP-EP list of samples you want to investigate.
• Pseudocolor Cy3/Cy5 (or F1/F2) ratio or Zdiff [RB] - display the Cy3 and Cy5 (or F1 and F2 for duplicate spotted grids) corresponding spots for the same genes as a pseudocolor array image as either a ratio (Cy3/Cy5) or Zdiff (Cy3-Cy5) if Zscore normalization is in effect. The high ratios or Zscores are red and low values are green. Values in the middle are black.
• Flicker HP-X & HP-Y Pseudoimages [RB] - toggle flickering the HP-X and HP-Y pseudograyscale intensity array images.
• Original image - display the original microarray 8-bit image (if available) for the current HP sample in a separate Web browser window.
• -----------------------------------------
• Use dual HP-X & HP-Y Pseudoimage [CB] - If have replicate spots (F1,F2), toggle the display between F1 data in left grids and F2 data in the right grids to mean HP-X data in the left grids and HP-Y in the right grids for side by side comparisons.
• "Scale pseudoarray image by 1/100 to zoom low-range values [CB] - rescales intensity and (Cy3+Cy5) and (HP-X + HY-Y) (Red-Yellow-Green) plots to so mid values are easier to visualize.
• SaveAs GIF - save the current microarray pseudo image as a full resolution GIF file specified by the user in a popup file browser window (stand-alone mode only).

The same data is shown in a variety of normalization and display formats.

2.4.4.1.1 Examples of microarray intensity data pseudoarray image

Figure 2.4.4.1.1.1 Pseudoarray intensity image of median normalized intensities of the current HP sample (C57B6 virgin 10 weeks from MGAP database). The graylevel scale on the left edge of the pseudoarray image indicates the spot intensity. All pseudoarray images have scales that vary depending on the type of pseudoarray being displayed.

Figure 2.4.4.1.1.2 Pseudoarray intensity image of Zscore normalized intensities of the current HP (C57B6 virgin 10 weeks from MGAP database).

Figure 2.4.4.1.1.3 Pseudoarray intensity image of ZscoreLog normalized intensities of the current HP (C57B6 virgin 10 weeks from MGAP database).

Figure 2.4.4.1.1.4 Pseudoarray intensity image of ZscoreLog normalized intensities of the dual HP-X and HY-Y individual samples. The Plot menu Show Microarray submenu toggle "Use dual HP-X & HP-Y samples" option is set. HP-X is a C57B6 pregnancy day 13 and HP-Y is a Stat5a (-,-) pregnancy day 13.

Figure 2.4.4.1.1.5 Pseudoarray intensity image of ZscoreLog normalized intensities of the dual HP-X and HY-Y sample 'sets'. The Plot menu Show Microarray submenu toggle "Use dual HP-X & HP-Y samples" option is set. The "Use HP-X & HP-Y 'sets' option in the Samples menu. HP-X is the mean of three 'C57B6 pregnancy day 13' and HP-Y is the mean of three 'Stat5a (-,-) pregnancy day 13'.

2.4.4.1.2 Example of microarray ratio or Zdiff data pseudocolor image

Figure 2.4.4.1.2.1 Pseudocolor array image of median normalized X/Y ratios. HP-X is C57B6 pregnancy day 13 and HP-Y is Stat5a (-,-) pregnancy day 13. Each spot's color represents the normalized X/Y ratio depending on Normalization mode. The color of the box is one of 9 colors representing the normalized expression ranges and assigned according to the table "Ratio mode".

Figure 2.4.4.1.2.2 Pseudoarray color image of normalized X/Y 'set' mean value ratios. Mean of three HP-X C57B5 pregnancy day 13 samples and mean of three HP-Y Stat5a (-,-) pregnancy day 13 samples. Each spot's color represents the normalized X/Y 'set' ratios depending on Normalization mode. The color of the box is one of 9 colors representing the normalized expression ranges and assigned according to the table "Ratio mode".

Figure 2.4.4.1.2.3 Pseudoarray color image of X-Y Zdiffs. HP-X is C57B6 pregnancy day 13 and HP-Y is Stat5a (-,-) pregnancy day 13. Each spot's color represents the normalized X-Y Zdiff depending using the Zdiff normalization mode. The color of the box is one of 9 colors representing the normalized expression ranges and assigned according to the table "Zdiff mode".

Figure 2.4.4.1.2.4 Pseudoarray color image of X-Y Zdiff of log data. HP-X C57B5 pregnancy day 13 sample and HP-Y Stat5a (-,-) pregnancy day 13 sample. Each spot's color represents the normalized X/Y ratio depending on ZdiffLog with StdDev normalization mode. The color of the box is one of 9 colors representing the normalized expression ranges and assigned according to the table "ZdiffLog mode".

Figure 2.4.4.1.2.5 Pseudoarray image showing color-coded p-values for t-test comparison of HP-X and HP-Y 'set' samples. The HP-X and HP-Y sets both have 2 samples each (more is obviously much better). The data was normalized using the Median and a spot intensity [SI1:SI2] data filter was applied to eliminate some of the noisy data. Each spot's color represents a p-value in the range indicated in the scale in the left edge of the image. Note that although all spots are assigned a p-Value, many may not be very significant because adequate preprocessing of the data (such as normalization, and low intensity spot removal, etc.). So use this display with care.

2.4.4.2 Scatter plots menu

  rSq=0.974, n=1728, X(mn+-sd)=(4.477+-7.845), Y(mn+-sd)=(12.379+-24.810)

If Cy3/Cy5 ratio data is being analyzed, then the "HP F1 vs F2 intensity" menu entry becomes

• HP Cy3 vs Cy5 intensity - display a scatter plot of Cy3 vs Cy5 intensity data of the current sample selected (i.e. being displayed).

The following figure illustrates some of the scatter plots and zoomed regions using the scroll bars on the horizontal and vertical axes.

A)

B)

C)

Figure 2.4.4.2 Scatter plot of HP-X and HP-Y single sample data. HP-X is C57B6 pregnancy day 13 and HP-Y is pactation day 1. A) An active scatter plot may be generated for the current HP-X and HP-Y samples filtered by "All named genes". B) similar plot for HP-X and HP-Y 'sets' of replicate samples (3 pregnancy and 4 lactation samples in the sets respectively). Clicking on a point in the plot sets the current gene. C) Zoomed up region (of B) at the bottom of the plot showing more detail and filtered by just "All named genes". Zooming is performed by adjusting the X or Y axes limits scroll bars. Note the points enclosed in magenta boxes indicate genes in the E.G.L. gene list.

Scatter plots of data from multiple channels on the same sample

A)

B)

C)

D)

Figure 2.4.4.2.1 Scatter plot of multiple channel data from a single sample. A) F1 Vs F2 data for a C57B6 pregnancy day 13 sample. B) Cy3 vs Cy5 data for a NCI mAdb mouse array sample. C) Scatter plot of individual Cy3 channels from two different ratio Cy3/Cy5 data hybridized samples. C) Scatter plot of individual Cy3 channel of HP-X compared with Cy3 channel of HP-Y for ratio Cy3/Cy5 data hybridized samples. D) Scatter plot of individual Cy3 channel of HP-X compared with Cy5 channel of HP-Y for ratio Cy3/Cy5 data hybridized samples.

2.4.4.3 Histogram plots menu

A)

B)

C)

Figure 2.4.4.3 Histogram plots. A) Ratio histogram of HP-X/HP-Y data with particular histogram bin selected with the constraint set to filter all genes > that bin. HP-X is 13 day pregnancy C57B6 and HP-Y is day 1 lactatation. The selected bin thresholds are then used in the Filter with the resulting Filtered genes shown in the array image. B) Zdiff histogram of HP-X - HP-Y 'sets' for same data as (A) but with the >< threshold constraint set to find genes outside of the symmetric histogram range. C) Intensity histogram of HP-X data filtered by [I1:I2] intensity range. As with ratio histograms, you can do additional filtering by selecting a particular histogram bin that is then used in the Filter. Filtering was disabled for the intensity histogram. To apply the filter, the "Don't re-Filter" button would be toggled to the "Re-Filter" state. The threshold constraints include: =, >, <, >, <>, and ><. Note that each time you click on the "Thr:" button, it cycles to the next option in the threshold constraints list.

2.4.4.4 Expression profile plots menu

You may interrogate a line corresponding to a particular HP sample in a EP plot by moving the mouse over the line and then selecting the line. This will cause the name of the HP, its intensity and CV to appear in the plot. If the Err check box is set, then the mean of the intensity is indicated by a short horizontal bar and the +- CV by red vertical error bars above and below the mean. If the plot style Line button is pressed, then the plot style is cycled between Line (vertical lines for each point), Circle (small circles at each point), and Curve (circles are connected). Pressing the button repeatedly cycles through: Line (i.e. vertical vars), Circle, or Curve (i.e. continuous curve of all samples). In the case of mean expression profiles used in K-means clustering, the standard deviation is used in place of the CV value. The various clustering methods have EP plots buttons. When they are invoked, the scrollable list of EP plots is sorted by the clustering method ordered list of genes. This enables you to view the data in the same order as that produced by the cluster analysis. If the zoom nnX button is pressed, then all of the plots are magnified by nn-fold to make low intensity plots more visible. Pressing the button repeatedly cycles through: 1X, 2X, 5X, 10X and 20X. It does not change the data itself. The Show HP names button pops up a numbered list of all HP entries used in the expression profile. If you are in stand-alone mode, a SaveAs GIF button will also be available for the EP overlay mode (Figure 2.4.4.4.1) or individual EP plot. This saves the current plot as a full resolution GIF file specified by the user in a popup file browser window.

The Expression profile plots submenu contains:

• Display a gene's expr. profile for HP-E - popup a window and display the expression profile of a gene when click on a spot in the Image or a point in the scatter plot.
• Display Filtered genes expr. profiles for HP-E - popup a scrollable montage of expression profiles of the list of Filtered genes.
• --------------------
• Use EP overlay else EP list [CB] - display Filtered genes as an overlay plot of expression profiles, else as a scrollable list of EP plots.

A)

B)

Figure 2.4.4.4 Expression profile plots. A) Individual expression profile plots may be created by clicking on any gene. Multiple instances may also be created. Here we show some of the presentation options for the 38 sample MGAP database. Error bars are computed for the standard error for that sample. There are three different plotting options: line, circle and curve. #1 is the default line plot with error bars. #2 is the line plot without the error bars but clicking on line 7 to find out which sample it is and what the intensity value is. #3 is the circle plot with error bars, and #4 is the curve plot without error bars. Window #5 shows the list of samples corresponding to the 38 points in the EP plots. B) List of EPplots of the oncogenes and proto-oncogenes in the database (set by the guesser with "onco" and "Set E.G.L." and the Edited Gene List Filter). The list would become scrollable if there were more than 10 profiles. Setting the current gene would scroll the list to the EPplot for the current gene.

A)

B)

C)

Figure 2.4.4.4.1 Expression profile plots. A) Scrollable list of EP plots of Filtered named genes centered at Carbonic anhydrase III. B) Overlay plot of all named Filtered genes. C) Overlay plot of all ONCO or PROTO-ONCO genes with the draw EGL option active so the graphs are drawn for these genes.

2.4.5 Cluster menu

Figure 2.4.5 Cluster Menu options. The hierarchical clustering option is being selected.

Use of clustering to find patterns of similar gene expression

There are many methods for doing clustering - each with advantages and disadvantages. We present three methods in MAExplorer and plan on adding a variety of more powerful methods through the MAEPlugin facility under development.

These methods may find genes belonging to particular clusters or genes that cluster well with particular genes. Gene clusters are sets of genes whose expression profiles are found to be similar according to a particular metric. We now define what we mean by "similar". The order list of hybridized samples used in computing the expression profiles are those in the HP-E list. MAExplorer has two different dissimilarity measures for C_ij: Euclidean distance LSQdist_ij and Pearson correlation coefficient r_ij. These are computed as follows and are tested against the cluster distance threshold (set by the slider in the preferences sliders). Let n= |HP-E|, the number of samples in the expression profile. We define similarity as (1.0 - normalized dissimilarity).

   LSQdistij = Sqrt( Sum ( D'hj - D'hi) **2 ) / n 
               h in HP-E
	       i,j in Filtered genes, i not j

Let,
   sumij = Sum( D'hj * D'hi ),
   mni = (1/n)Sum( D'hi ),
   mnj = (1/n)Sum( D'hj ),
   sumSqi = Sum( D'hi *  D'hi ),
   sumSqj = Sum( D'hj *  D'hj ),
  
then,
         [sumij - n*(mni * mnj)] 
   rij = --------------------------------------------------------
         [Sqrt(sumSqi - n*n*mni*mni) * Sqrt(sumSqj - n*n*mnj*mnj)]
               h in HP-E
	       i,j in Filtered genes, i not j

The Cluster plots submenu contains a number of clustering methods. Pressing the Escape key during a long cluster operation will abort the operation. If you are in stand-alone mode using the ClusterGram, a SaveAs GIF button will also be available for saving the current plot as a full resolution GIF file specified by the user in a popup file browser window.:

• Cluster genes with expression profiles similar to current gene [RB] - click on gene in image to find other genes with similar HP-E expression profiles whose cluster distance is less than the cluster distance threshold. The larger the blue box, the higher the similarity.
• Cluster counts of similar Filtered genes by expression profiles [RB] - draw blue circles around filtered genes indicating the number of other genes whose cluster similarity is less than the cluster distance threshold. The larger the circle, the more similar genes were found. Clicking on a gene switches to the above mode.
• K-means clustering of gene expression profiles [RB] - draw magenta circles around the N primary-node gene clusters representing the gene closest to representing the center of the cluster. Each of the nodes is a maximum distance from all other nodes in the recursive definition of nodes. N is determined by the State Scroller "# of Clusters". Changing N will recompute the clusters. It then pops up a scrollable text window with the clusters and indicates which genes belong to it. If you select the EP plot button, it will draw the expression profiles for the clustered genes. The Mn-Cluster-Report button will generate report for all genes sorted by K-means cluster. Summaries can be generated using the Mean EP plot and Mn-Cluster-Report buttons. The SaveAs GeneSets button saves all of the clusters as named Gene Sets ("Cluster #1", "Cluster #2", etc). If you change the filter or current gene, you should explicitly use the Recompute Clusters button to regenerate the new set of clustered genes. When you recompute the K-means clusters, it uses the current gene as the initial node.
• Hierarchical clustering of expression profiles - this computes the hierarchical clustering of the expression profiles (normalized by HP-X sample data for each gene) of Filtered genes. The hierarchical clusters are displayed in an ordered gene clustergram and optional dendrogram. Sub-regions of the clustergram may be explored in more detail using the EP-subset plot button, or a report of the ordered genes can be created using the ClustGram Report Note: you may add (remove) genes you select from the Clustergram to the E.G.L. by holding the Control(Shift) key while clicking on the gene name.
• S.O.M. gene clusters by expr profiles [RB] - [Future MAEPlugin]
• Multi-Dimensional Scaling of genes by expr profiles [RB] - [Future MAEPlugin]
• Multi-Dimensional Scaling of genes by exprprofiles [RB] - [Future MAEPlugin]
• Clusters of (HP-E) samples as fct of Filtered genes [RB] - [Future]
• --------------------
• Use correlation-coefficient else Euclidian-distance [CB] - use the (1.0 - correlation coefficient) as the distance metric instead of the default Euclidean distance.
• Scale EP vector by max magnitude prior to clustering [CB] - scale each sample in the EP by the max magnitude for all sample values in the EP.
• Normalize by HP-X sample else HP max intensities [CB] - normalize data by the corresponding HP-X sample data for each gene or the maximum raw intensity for each HP in the expression profile.
• Use median instead of mean for K-means clustering [CB] - use the clustering (see (Bickel, 2001)).

The Hierarchical Cluster plots submenu contains:

• Display ClusterGram of gene expr profiles [CB] - compute the hierarchical clustering of the expression profiles (normalized by HP-X sample data for each gene) of Filtered genes. Then display the hierarchical clusters in an ordered gene clustergram and optional dendrogram when the dendrogram checkbox is selected. Expression profile plots of the clustergram may be explored in more detail using the EP plot button that generates a scrollable list of all EP plots ordered by the same order as the clustergram. A full report of the ordered genes expression profiles may be created using the ClustGram Report button.
• --------------------
• Use avg-arithmetic-linkage [RB] - set the hierarchical clustering linkage method to the average arithmetic linkage of sub-clusters.[ Future]
• Use avg-centroid-linkage [RB] - set the hierarchical clustering linkage method to average centroid linkage of sub-clusters (default).
• Use next-min-linkage [RB] - set the hierarchical clustering linkage method to the next minimum distance sub-cluster linkage in random order.
• Use cluster-distance matrix cache [CB] - if you do not have enough memory for clustering large gene sets, disable the cache. It will take MUCH longer without the cache. When clustering, if there is not enough memory available for the cache, it will warn you and suggest you either reduce the number of genes being clustered or use a computer with more memory.)
• Use short else float cluster-distance matrix cache [CB] - if there is not enough memory for the set of genes you wish to cluster and you still want to use the cache, you can use 16-bit (i.e. short) data instead of the 32-bit (i.e. float) data. The results will be less precise.
• Use un-weighted else weighted average [CB] - set the hierarchical clustering vector averaging to un-weighted (the default weights it by the number of genes in that sub-cluster). Otherwise using weighted gives equal (0.50) weighting to each sub-cluster.

Handling of hierarchical clustering of large numbers of genes - problem with slow response

1. If you determine that your computer is "paging", use a computer with more memory. The currently distributed stand-alone version will use up to 256 Mbytes of chip memory.
2. If that is not possible, reduce the number of genes being clustered. Even if you have enough memory, the computation is still high if N is large.
3. Set the Use short else float cluster-distance matrix cache option in the cluster plot menu. This reduces the memory requirements for the distance matrix by 1/2.

2.4.5.1 Cluster genes with expression profiles similar to current gene

A)

B)

Figure 2.4.5.1 Similar genes clustered to the current gene. This method finds all genes that are similar to the current gene as those defined by their distance between expression profiles being less than the threshold set by the user. Each gene that passes the cluster distance threshold test is indicated in the image with a blue square where the size of the square is proportional to its similarity. This data is from the 38 samples in the MGAP database containing duplicated spots. A) Main windows with popup cluster similarity report and cluster distance threshold slider. B) Scrollable list of EPplots of similar genes with the red error bars indicating the variation for duplicated spots for each HP sample. The Err checkbox may turn the error bar overlays on and off.

2.4.5.2 Cluster counts of similar filtered genes by expression profiles

Figure 2.4.5.2 Display of cluster counts for all genes less than the cluster threshold from MGAP 38 sample database. The algorithm counts the number of similar genes for each Filtered gene and draws a blue circle whose size is proportional to the number of genes similar to that gene. That is why there are a larger number of the larger circles.

2.4.5.3 K-means clustering' gene expression profiles for filtered genes

A)

B)

C)

Figure 2.4.5.3 Genes clustered using the K-means cluster method. A) Using the current gene as the initial cluster, MAExplorer finds N orthogonal clusters assigning the set of filtered genes to these clusters using the HP-E expression profiles. All genes are iteratively assigned to these clusters. Genes belonging to the current cluster are labeled with a green cluster number both in the array and in the scatter plot. The slider determines the number of clusters (set to 6 here). A 2D scatter plot shows the genes belonging to cluster 6. The K-means cluster report on the right contains a sorted list of the genes in each cluster and has buttons to generate EP plots and reports as well as summary mean EP plots (shown) and mean cluster reports. The detailed list is shown below. B) Part of the scrollable EP plots for this data showing genes belonging to both clusters #5 and #6. C) The mean EP plots for the 6 clusters.

 
Cluster report for 6 K-means clusters with 141 genes being clustered.
The seed gene is [1248564] Jun-B oncogene.

Clone ID  Similarity      Cluster-#  Distance-to-cluster  Gene-Name
--------  --------------  ---------  -------------------  ----------------

1248411   **************  1          Cluster [26 genes] in cluster [distNext: 1.035] wiCdist:mn+-sd=1.223+-0.453 CV=0.371  Calpactin I light chain
1381592   **********      1          0.448  Surfeit gene 4
1247956   *********       1          0.706  Protein kinase, cAMP dependent, catalytic, beta
1381836   ********        1          0.761  Prohibitin
1382325   ********        1          0.771  M.musculus mRNA for C1D protein
1248270   ********        1          0.775  Seven in absentia 1A
1247716   ********        1          0.794  Lipoprotein lipase
1248184   ********        1          0.847  Mus musculus bromodomain-containing protein BP75 mRNA, complete cds
1248564   *******         1          0.864  Jun-B oncogene
1382667   *******         1          0.888  SERINE/THREONINE PROTEIN PHOSPHATASE PP2A-BETA, CATALYTIC SUBUNIT
1382561   *******         1          0.931  Mus musculus GTP-specific succinyl-CoA synthetase beta subunit (Scs) mRNA, partial cds
1248089   ******          1          1.013  M.musculus RPS3a gene
1247780   ******          1          1.088  Proprotein convertase subtilisin/kexin type 7
1247557   ******          1          1.104  M.musculus L28 mRNA for ribosomal protein L28
1248321   *****           1          1.278  Decay accelerating factor 1
1382751   ****            1          1.311  Clusterin
1382007   ****            1          1.357  Murine mRNA with homology to yeast L29 ribosomal protein gene
1382074   ****            1          1.390  Orosomucoid 1
1381963   ****            1          1.417  M.musculus mRNA for ribosomal protein L36
1248278   **              1          1.658  HISTONE H3.3
1247630   **              1          1.675  Procollagen, type I, alpha 2
1247865   *               1          1.837  Mouse beta-D-galactosidase fusion protein mRNA, complete cds
1382236   *               1          1.85  Caspase 7
1247833                   1          1.882  Mus musculus radio-resistance/chemo-resistance/cell cycle checkpoint control protein (Rad9) mRNA, complete cds
1248535                   1          1.953  M.musculus mRNA for selenoprotein P
1247702                   1          2.157  Cytochrome C oxidase, subunit Va
1382282   **************  2          Cluster [13 genes] in cluster [distNext: 24.199] wiCdist:mn+-sd=16.184+-6.667 CV=0.412  Max interacting protein 1
1382159   **********      2          9.086  TRANSPLANTATION ANTIGEN P35B
1247854   *********       2          11.002  Prolyl 4-hydroxylase, beta polypeptide
1247970   ********        2          11.786  Mouse mRNA for osteoblast specific factor 2 (OSF-2)
1381663   ********        2          12.948  Mus musculus vacuolar adenosine triphosphatase subunit A gene, complete cds
1382100   ********        2          13.34  T-complex protein 1, related sequence 1
1248366   ********        2          13.541  Mus musculus cytochrome c oxidase subunit VIIa-L precursor (Cox7al) mRNA, nuclear gene encoding mitochondrial protein, complete cds
1247568   ********        2          13.762  Cathepsin D
1247872   *******         2          14.015  Mus musculus endothelial monocyte-activating polypeptide I mRNA, complete cds
1382333   *******         2          14.065  Stromal cell derived factor 5
1382008   *******         2          15.985  Mus musculus FK-506 binding protein homolog (SAM11) mRNA, complete cds
1247724   ****            2          21.964  Glutathione-S-transferase, alpha 3
1247846                   2          34.704  House mouse; Musculus domesticus kidney mRNA for Phosphatidic acid phosphatase, complete cds
1247945   **************  3          Cluster [22 genes] in cluster [distNext: 11.979] wiCdist:mn+-sd=7.559+-3.347 CV=0.443  Mus musculus mRNA for DEDD protein
1247797   **********      3          4.159  Mus musculus Btk locus, alpha-D-galactosidase A (Ags), ribosomal protein (L44L), and Bruton's tyrosine kinase (Btk) genes, complete cds
1382087   **********      3          4.494  Cell division cycle 42
1247539   **********      3          4.511  EST
1248212   **********      3          5.009  Murine mRNA for integrin beta subunit
1248470   **********      3          5.044  EST
1247521   *********       3          5.299  Mus musculus mRNA for peroxisomal integral membrane protein PMP34
1381808   *********       3          5.924  Mus musculus UDP-GalNAc:polypeptide N-acetylgalactosaminyltransferase-T3 mRNA, complete cds
1381970   *********       3          6.285  Mus musculus thioredoxin mRNA, nuclear gene encoding mitochondrial protein, complete cds
1382168   *********       3          6.343  N-terminal Asn amidase
1382704   *********       3          6.36  Mus musculus N-myristoyltransferase 1 mRNA, complete cds
1248548   *********       3          6.378  Mus musculus WDR protein mRNA, complete cds
1247564   ********        3          6.652  Erythrocyte protein band 7.2
1248588   ********        3          6.67  M.musculus BAP31 mRNA
1247541   ********        3          6.690  Apolipoprotein D
1248462   ********        3          7.322  Sterol O-acyltransferase 1
1248462   ********        3          7.42  Sterol O-acyltransferase 1
1248521   ******          3          9.121  Mus domesticus nuclear binding factor NF2d9 mRNA, complete cds
1382212   ******          3          10.137  Thyroid autoantigen 70 kDa
1382270   *****           3          10.529  Voltage-dependent anion channel 2
1248152   *****           3          10.541  M. musculus mRNA for MAP kinase-activated protein kinase 2
1247678                   3          19.431  Casein alpha
1247543   **************  4          Cluster [44 genes] in cluster [distNext: 1.035] wiCdist:mn+-sd=0.439+-0.266 CV=0.606  RAS-related C3 botulinum substrate 1
1381923   ************    4          0.158  Prolyl 4-hydroxylase, beta polypeptide
1382052   ************    4          0.209  Trans-acting transcription factor 1
1247882   ***********     4          0.237  Mus musculus AMP activated protein kinase mRNA, complete cds
1248099   ***********     4          0.246  Mus musculus mitogen-responsive 96 kDa phosphoprotein p96 mRNA, alternatively spliced p67 mRNA, and alternatively spliced p93 mRNA, complete cds
1248351   ***********     4          0.251  Abl-interactor 1
1247540   ***********     4          0.255  Mus musculus mRNA for ZIP-kinase, complete cds
1248316   ***********     4          0.26  Mus musculus proteasome alpha7/C8 subunit mRNA, complete cds
1382671   ***********     4          0.264  Mouse MA-3 (apoptosis-related gene) mRNA, complete cds
1382014   ***********     4          0.277  Transcription elongation factor B (SIII), polypeptide 1 (15 kDa),-like
1247885   ***********     4          0.289  Mus musculus mRNA for ryudocan core protein, complete cds
1248294   ***********     4          0.292  Mus musculus thioredoxin-related protein mRNA, complete cds
1382066   ***********     4          0.306  Inhibitor of DNA binding 2
1248597   ***********     4          0.307  Lipocortin 1
1248591   ***********     4          0.324  Interferon beta, fibroblast
1248445   **********      4          0.333  Mus musculus beta prime coatomer protein mRNA, partial cds
1247775   **********      4          0.34  House mouse; Musculus domesticus male brain mRNA for ARF1, complete cds
1382750   **********      4          0.340  Thymoma viral proto-oncogene
1247905   **********      4          0.341  Monokine induced by gamma interferon
1381668   **********      4          0.351  Mus musculus mitogen-activated protein kinase-activated protein kinase mRNA, complete cds
1381811   **********      4          0.356  Protein tyrosine phosphatase, receptor type, D
1382031   **********      4          0.358  Protease (prosome, macropain) 28 subunit, beta
1248345   **********      4          0.363  Mus musculus alpha-methylacyl-CoA racemase mRNA, complete cds
1382555   **********      4          0.364  Lysosomal membrane glycoprotein 1
1247820   **********      4          0.367  Tight junction protein 1
1247598   **********      4          0.374  Retinoblastoma 1
1247595   **********      4          0.378  PROBABLE CALCIUM-BINDING PROTEIN PMP41
1381928   **********      4          0.379  Mus musculus MRJ (Mrj) mRNA, complete cds
1248196   **********      4          0.399  Max protein
1381691   **********      4          0.423  SRY-box containing gene 17
1248225   **********      4          0.434  Mus musculus heat shock transcription factor 1 (Hsf1) gene, partial cds
1248084   **********      4          0.442  Mus musculus Supl15h gene
1247941   *********       4          0.453  Fibroblast growth factor inducible 14
1381623   *********       4          0.468  Stearoyl-coenzyme A desaturase 1
1248202   *********       4          0.473  Mouse mRNA for PAP-1, complete cds
1382115   *********       4          0.512  GLUTATHIONE S-TRANSFERASE GT8.7
1382044   *********       4          0.515  Cartilage derived retinoic acid sensitive protein
1381636   ********        4          0.567  Lymphotoxin B
1381920   ********        4          0.569  Mus musculus mRNA for NEFA protein, complete cds
1247757   ********        4          0.596  Granzyme B
1382094   ********        4          0.609  High mobility group protein 1
1247545   ********        4          0.638  Carbon catabolite repression 4 homolog (S. cerevisiae)
1247607   ***             4          1.188  POLYADENYLATE-BINDING PROTEIN
1247727                   4          1.667  Malate dehydrogenase, mitochondrial
1248244   **************  5          Cluster [19 genes] in cluster [distNext: 3.473] wiCdist:mn+-sd=4.273+-2.059 CV=0.482  CD80 antigen
1248534   **********      5          1.648  Carbonyl reductase
1247764   **********      5          1.776  H-2 CLASS II HISTOCOMPATIBILITY ANTIGEN, GAMMA CHAIN
1381933   *********       5          2.345  Mouse rpS17 mRNA for ribosomal protein S17, complete cds
1381616   *********       5          2.42  Mus musculus oral tumor suppressor homolog (Doc-1) mRNA, partial cds
1248232   *********       5          2.486  Mus musculus putative glycogen storage disease type 1b protein mRNA, complete cds
1382644   ********        5          2.717  Cyclin G
1248125   ********        5          2.791  Histocompatibility 2, class II, locus Mb2
1247799   ********        5          2.869  Mus musculus signal recognition particle receptor beta subunit mRNA, complete cds
1247708   ********        5          3.024  Ephrin A1
1247932   ******          5          4.235  Mus musculus (clone: pMAT1) mRNA, complete cds
1382515   *****           5          4.668  ATPase, Na+/K+ beta 3 polypeptide
1248586   *****           5          4.838  Mus musculus viral envelope like protein (G7e) gene, complete cds
1248198   ***             5          5.874  Mus musculus D9 splice variant 2 mRNA, complete cds
1381623   **              5          6.224  Stearoyl-coenzyme A desaturase 1
1382086   *               5          6.885  Mus musculus (strain C57Bl/6) mRNA sequence
1247887   *               5          7.014  Mouse chromosome 6 BAC-284H12 (Research Genetics mouse BAC library) complete sequence
1247886                   5          7.810  Cut (Drosophila)-like 1
1248303                   5          8.094  Lipopolysaccharide response
1247621   **************  6          Cluster [17 genes] in cluster [distNext: 19.157] wiCdist:mn+-sd=12.410+-3.024 CV=0.244  Mus musculus Lsc (lsc) oncogene mRNA, complete cds
1248050   *******         6          7.407  Mus musculus C57BL/6J ribosomal protein S28 mRNA, complete cds
1247698   *******         6          7.571  Adipocyte protein aP2
1248240   *****           6          9.198  Mus musculus mRNA, complete cds
1247862   ****            6          9.844  Mus musculus Nmi mRNA, complete cds
1382162   ****            6          10.330  CAMP responsive element modulator
1248398   ***             6          11.007  Mouse mRNA for ribosomal protein S12
1248281   ***             6          11.143  M.musculus mRNA for histone H3.3A
1247852   ***             6          11.576  Twist gene homolog, (Drosophila)
1381991   **              6          12.809  Prolyl 4-hydroxylase, beta polypeptide
1382753   **              6          13.019  Mus musculus cleavage and polyadenylation specificity factor (MCPSF) mRNA, complete cds
1248368   *               6          13.639  Mus musculus ribosomal protein S26 (RPS26) mRNA, complete cds
1247639   *               6          13.692  SRY-box containing gene 4
1248435                   6          14.262  Thymus cell antigen 1, theta
1247961                   6          14.75  ATP SYNTHASE ALPHA CHAIN, MITOCHONDRIAL PRECURSOR
1248344                   6          15.217  Gut enriched Kruppel-like factor
1382234                   6          16.351  CD8 antigen, beta chain

We call the genes closest to the "center" of the K clusters primary genes and they are reported with additional information. The "Cluster [# genes]" entries in the distance-to-cluster fields indicates that these genes are the center of the clusters (i.e. primary genes). The distNext is the distance from this cluster center to the next nearest K-means cluster center. The number of clusters N (6 in this example) is set in the popup state scroller. If you change the value of N, it will recompute the clusters and the primary genes.

It draws magenta circles around the primary genes in the microarray and the cluster number to the right of the circle. The size of a circle corresponds to the number of genes clustered with that circle. If you click on a gene belonging to any cluster, it defines that cluster as the "current cluster". It will change the labels of the subset of genes that belong to the current gene from red (white) circle to a green (yellow) cluster number of the current cluster in the intensity (ratio) pseudoarray image. In addition, the 'edited gene list' is set to the subset of genes that belong to the current cluster. If you are also displaying a scatter plot, genes in the current cluster have their red '+' characters changed to the cluster number.

You can click on that gene in the array image to determine its identity. You may also popup an ordered (same as the above report) plot of the clusters expression profiles by clicking on the EP plot button. You may plot the mean expression profiles of the N clusters using the Mean EP plot button. You may generate a report of all of the clustered genes or of the mean clusters using the Cluster-Report or Mn-Cluster-Report buttons respectively. If you change the Filter conditions, you may recompute the clusters using the Recompute Clusters button. Closing the text window will remove the magenta circles. If you selected the current cluster, the genes that belong to it will still be available in the 'edited gene list' for making reports, saving as a gene subset or for additional gene filtering. If you press the SaveAs GeneSets button, then K gene sets are created with the names "Cluster#1", "Cluster#2", ..., "Cluster#K". You can then save or rename the clusters you want and delete the rest. If you press the ClusterGram button, it displays the gene sets in a cluster gram order the same way as the cluster report.

2.4.5.4 Hierarchical clustering of expression profiles

Clustering is represented by a binary tree and is visualized as an ordered gene clustergram and optional dendrogram sub-plot. This is similar to the methods of (DeRisi, 1996), (Eisen, 1998), and (White, 1999). Currently, MAExplorer does 1-way clustering - not the 2-way clustering of (Weinstein, 1998) and (Eisen, 1998). Each row of the clustergram represents a gene and each column represents a HP in the HP-E list of samples. Each box in a row represents the normalized expression of that gene for the HP represented in that column. The color of the box is one of 9 colors representing the normalized expression ranges and assigned according to the following table:

A)

B)

C)

Figure 2.4.5.4 Hierarchical clustering clustergram of genes filtered by ratio histogram bins for 19 samples from the MGAP data set. The hybridized samples are drawn as colored boxes in the 19 columns. Rows of boxes correspond to gene expression profiles. In A), the set of all genes and ESTs was filtered by the CV filter set to 0.387 and the normalization was the Zscore. The gene "Mus musculus D9 spice variant 2 mRNA, complete cds" was selected as the current gene in the clustergram. Data for this gene and the selected HP column is indicated at the top of the clustergram. The list of the 19 samples is shown on the left. B) Details of clustergram and dendrogram are shown where the user had selected a cluster distance threshold at "Mouse mRNA for mitochondrial cytochrome c oxidase subunit Vb" in the dendrogram part of the plot (zoomed by 2X). This selection draws all parts of the dendrogram tree that are less than this distance are drawn in red. C) shows the manual selection of genes from the ClusterGram or Dendrogram by clicking on the genes names you wish to capture in the Edited Gene List (EGL) while the Control key is pressed. The zoomed subregion shows three genes in the same cluster that were selected (magenta stars in the right edge of the ClusterGram).

2.4.6 Report menu

Figure 2.4.6 Reports menu. You may create either dynamic or tab-delimited text reports of either Samples or of subsets of genes.

These may be presented as interactive dynamic tables as well as scrollable text windows capable of being exported to Excel. If Web DB access is enabled, clicking on an entry will bring up a Web browser with access to GenBank data. If the report contains Clone ID as one of the fields, you can click on it to have it define that gene as the current gene and highlight it in the microarray image or scatter plot (if it is being used). The reports are divided into two types - those dealing with lists of arrays (i.e. the sample experimental condition) and those dealing with lists of genes.

The Report menu includes:

• Array reports - show tables of sample array information.
• Gene reports - show tables of sets of genes information.
• --------------------
• Table format - generates the table as 'Spreadsheet', 'Tab delimited', or 'Name=Value list'.
• Table font size - changes the report font size from default 10 point.

2.4.6.1 Array report menu - hybridized samples global data

  rSq=0.748, n=1656, HP:1(mn+-sd)=(28991+-19564), HP:2(mn+-sd)=(5044+-9766)

A)

B)

C)

Figure 2.4.6.1 Hybridized samples dynamic Report windows. A) Samples Info report. B) Sample Web links. Clicking on a blue hypertext link brings up the corresponding genomic Web database entry in a separate Web browser window if the Web access is enabled. The tab-delimited version of the same reports (not shown) may be cut and then pasted into other programs such as an Excel spreadsheet. C) HP vs HP correlation table on genes passing the data Filter for all samples in the HP-=E list.

2.4.6.2 Gene reports menu

2.4.6.2.1 Filtered gene reports menu

A)

B)

Figure 2.4.6.2 Gene Report windows of 50 named genes with highest HP-X/HP-Y 'set' ratios. A) Dynamic gene report of 50 genes with highest HP-X/HP-Y 'set' ratios. A similar report may be generated for the lowest ratios or for single HP-X/HP-Y samples. This type of report may be generated for the highest or lowest Zdiff values when the Zscore normalizations are used. Clicking on a blue hypertext link brings up the corresponding genomic Web database entry in a separate Web browser window if the Web access is enabled. It also sets the current gene to the gene for that row. B) The tab-delimited version of the same report may be cut and then pasted into other programs such as an Excel spreadsheet.

2.4.6.3 Table format menu

2.4.6.4 Table font size menu

2.5 View menu

Figure 2.5 View Menu options. These are divided into various options for modifying the presentation as well as recording activity such as the messages or history popup scrollable log windows.

• Show 'Edited Gene List' [CB] - toggle showing the EGL as magenta boxes in the pseudoarray image. If enabled, genes set by manual selection or as the result of some filtering operations (see Edit menu, the Filter and Clustering)
• ----------------------
• Enable display current gene in GenBank Web Browser [RB] - when click on a gene in microarray image or scatter plot to access NCBI data. It will use RefSeqID data if available.
• Enable display current gene in dbEST Web Browser [RB] - when click on a gene in microarray image or scatter plot to access NCBI data.
• Enable display current gene in Unigene Web Browser [RB] - when click on a gene in microarray image or scatter plot to access NCBI data.
• Enable display current gene in mAdb Web Browser [RB] - when click on a gene in microarray image or scatter plot to access NCI/CIT data.
• Enable display current gene in LocusLink Web Browser [RB] - when click on a gene in microarray image or scatter plot to access NCBI LocusLink data (by Locus ID if available, and GenBank ID if it] is not).
• Enable display current gene in OMIM Web Browser [RB] - when click on a gene in microarray image or scatter plot to access NCBI OMIM data by OMIM ID if available.
• ----------------------
• Show Filtered spots in array [CB] - if on, show Filtered genes in the microarray image using circle overlays just outside of each spot. Genes not passing the Filter have no circle.
• Gang F1-F2 scrolling [CB] - toggle gang scrolling. If on, gang measure F1 and F2 when click on one of them. If off, only measure only the spot that you click on.
• Show mouse-over info [CB] - toggle mouse over. If enabled, report HP or gene details when the mouse is moved over the sample names or spots in the array image or genes in scatterplot or expression profile overlay plot.
• Presentation view mode [CB] - toggle presentation mode. If enabled, increase the fonts to 12pt and draw darker circles, squares, and "+" so the display details are easier to see when projecting the display or making slides.
• Color scheme (red-green) or dichromasy [CB] - toggle between two color schemes red-green and orange-blue for dichromasy which may be easier for some people.
• Show log of messages [CB] - toggle message logging mode. If on, it pops up a scrollable log of all messages to the three line status area. This is useful for recording measurements and other activity. The messages may be saved in log file.
• Show log of command history [CB] - toggle command history logging mode. If enabled, it pops up a scrollable history of all commands issued to MAExplorer. The commands are automatically numbered. This is useful for recording what steps you took during an analysis. The history may be saved in log file.

A)

B)

C)

Figure 2.5. Popup genomic browser database page. A) The UniGene Web page pops up in a new Web browser window when the user clicks on a gene in the array image, 2D scatter plot or Report and the view is set to "Display current gene in Unigene Web Browser" toggle was enabled in the View menu. The current gene was "Jun-B oncogene". Alternatively, the B) mAdb Gene DB may be selected - as well as GenBank or dbEST genomic databases. C) Alternatively, data from the NCBI LocusLink database may be accessed if either the GenBank ID or LocusID is available.

Figure 2.5.2 Examples of messages and command history popup log windows. Measurements and other activity are shown in more detail in the messages window whereas the command history indicates commands (numbered in the order they are executed) in the command history window. Data from either of these windows may be saved in text log files.

2.6 Plugins menu

Figure 2.6 MAEPlugins paradigm. If you have a MAEPlugin .jar file, then it may be specified using the "Load plugin" command. When you invoke the command from the menus (or other methods), it accesses data from the current MAExplorer database it may need from the Open Java API.

• Load plugin - popup a browser to load a MAEPlugin jar file
• Unload plugin(s) - popup a selector to pick MAEPlugins to unload
• -------------------------------------
• RLO methods - list of executable R methods (R LayOuts). These may be added by the user using the RtestPlugin. An RLO analysis allows you to export data from MAExplorer, execute it with the R program, and import the R results back into MAExplorer. [This is under development and is alpha-level.]
• Save RLO reports in time-stamped Report/ folder [CB] - puts files generated by R from successive executions of the same RLO into separate sub-folders in the Report/ folder with names "RLOname-YYMMDD-HHMMSS/" to peep the data separate.
• -------------------------------------
• plugin #1 - 1st MAEPlugin added
• plugin #2 - 2nd MAEPlugin added
• . . .
• plugin #n - n'th MAEPlugin added

RLO methods menu

The Save RLO reports in time-stamped Report/ folder [CB] options puts files generated by R from successive executions of the same RLO into separate sub-folders in the Report/ folder with names "RLOname-YYMMDD-HHMMSS/" to peep the data separate. This is useful when you want to compare results from the same RLO method but with different MAExplorer preprocessing.

You may download the latest versions of all plugins using the (File | Update Plugins from maexplorer.sourceforge.net) menu command. Similarly, you can update your versions of the RLO methods using (File | Update RLO methods from maexplorer.sourceforge.net

2.6.1 Example of using a Plugin

Figure 2.6.1 Loading a MAEPlugin from your file system using the Load Plugins command in the Plugins pull down menu. If you have a plugin .jar or .class file, it may be specified using the "Load plugin" command. This pops up a file browser to let you specify the plugin file.

Figure 2.6.2 Executing the new command previously loaded in the Plugin menu. Selecting the new "Show List Active Filters" command that now appears in the Plugins menu invokes the plugin. This pops up a report shown in the next figure.

Figure 2.6.3 Popup window from executing the MAEPlugin. This plugin gives a full report on the data Filter status in a new pop up window.

Figure 2.6.2 Plugins menu - executing a previously loaded plugin. Plugins that do not go into particular MAExplorer submenus go into the Plugins menu. Selecting the command will invoke that MAEPlugin.

2.7 Help menu

The Help menu includes:

• Home page - MAExplorer home page on SourceForge.net
• Introduction - to MAExplorer
• Overview - overview of MAExplorer capabilities
• Short tutorial - simple tutorial of things to try in MAExplorer (Appendix A)
• Advanced tutorial - advanced tutorial of things to try in MAExplorer (Appendix B)
• Menu summary - short summary of the MAExplorer menus.
• Reference manual - this Reference Manual document
• MAEPlugins - plugin home page for MAExplorer
• Intro to data mining - for MAExplorer, (Chapter 3)
• Glossary - glossary of terms used in MAExplorer Reference Manual
• Index - Index of terms used in MAExplorer Reference Manual
• -------------------------------------

• Database-specific help menu entries - entries defined for a particular database (see below)

• -------------------------------------
• About - show MAExplorer authors, version and revision date

2.7.1 Adding custom help links to your database to the Help menu

• List of hybridized samples - available from MGAP.
• MGAP animal models - animal models used in MGAP.
• MGAP home page - Mammary Genome Anatomy Program

3. Exploratory Data Analysis - Introduction to Data Mining

This section briefly addresses some of the issues you need to consider. However, a full discussion of the issues involved is beyond the scope of this manual. These issues are covered in other more focused statistical methods literature and you might also address them in consultation with biostatisticians. The Internet has vast resources for microarrays. A few to get you started might include: a microarray citation electronic library http://arrayit.com/e-library/, the National Library of Medicine PubMed journal search engine, a general microarray Listserv GENE-ARRAYS@ITSSRV1.UCSF.EDU. The MGED group (Brazma, 2001) has published the MIAME standard which specifies (Minimum Information About a Microarray Experiment). This information is useful in doing an analysis. Also try searching using general Internet search engines. There are a number of public microarray data repositories. One that we find useful is NCBI's GEO (Gene Expression Omnibus), that contains array data and MIAME compliant information about the arrays.

• Data mining is a pattern discovery activity - use all the tools you have.
• It is open-ended because of the variety of ways data may be partitioned, normalized, pre-filtered, clustered, and viewed. Patterns that are apparent in one view may not be apparent in another.
• When data-mining microarray data, look at correlated genes from the point of view of what relationships might be interesting from a biological view by characterizing genes that cluster together using additional information. I.e. check out the results with various NCBI and PubMed database searches on the resulting genes, by designing other lab experiments to better uncover the cause and effect, etc. Remember correlation does not imply cause and effect.

Organization of Sections in this Chapter

3.1 Objectives in data mining, discovery and analysis

A good and appropriate experimental design (i.e. the design and setting up of experiments to subsequently be analyzed) is critical for resolving significant differences in gene expression between experimental conditions. We touch on some of the issues here. (Simon, 2001), (Dudoit,2000), and Kerr and Churchill (2001a, 2001b) discuss some of the issues of experimental design for microarrays. We do not currently implement the Kerr-Churchill method. However, some of the issues involved in experimental design based on the types of arrays are discussed in Section 3.1.1 for (Cy3/Cy5)-labeled as well as ³³P-labeled samples.

If users are comparing two different types of samples, the analysis would be different than if they were comparing an ordered sequence of samples (e.g. time series, cell cycle, dose-response, tumor-stage, etc.). MAExplorer gives users the ability to:

1. Organize their experiments by sample characteristics allowing them to perform a variety of mining analyses comparing gene expression patterns between sets of different samples or comparing a single sample within an array's set of genes.
2. Explore, compare, and record these analyses and share the results or intermediate data with other investigators.
3. Use graphical direct-manipulation combined with statistical methods, clustering and spreadsheet techniques to gain different insights into the structure of the data.
4. Access public Internet genomic databases for particular genes that are found to be interesting.

Briefly, data mining is the discovery of potentially interesting patterns in the data that were previously unknown. One approaches the analysis of a set of data with minimal expectations. However, some idea of what you are interested in helps focus the search. But beware of the trap of mining the data until you get the results you hope for. The following figure helps illustrate this process.

Figure 3.1 Flow chart of a typical data mining session. The user makes some initial decisions on the experimental design such as which hybridized samples to compare, the type and numbers of replicates. They then make initial guesses as to the normalization method to use, and the gene subset (the gene class) to concentrate on when setting the data filter. The data is viewed in various modalities to get a feeling for its inherent dynamic range and where interesting outliers might appear. Clustering and plots helps bring these differences into view. The results are then evaluated and either the process is finished or the views are refined by adjusting data normalization and filter parameters, data subsets to be investigated, clustering methods, plots etc. and the process repeated until the user is able to see the differences between gene subsets more clearly or no significant differences appear to be found.

Obviously, this approach is a first approximation to what is eventually required. But it does capture the flavor of the data-mining process. Typically the user would refine the search using variations of the data filters and might contrast (using gene sets and hybridized sample condition lists operations) results found under one set of conditions with those found under another set of conditions.

Recording the analysis steps during your data mining session - command history

3.1.1 Some experimental design issues of microarray experiments

Proper experimental design of microarray experiments is critical to successful use of microarray data. Several recent reports discuss some of the key issues involved in various aspects of statistical analysis of microarrays: (Radmacher, 2001), (McShane, 2001), (Korn, 2001), (Simon, 2001), (Dudoit,2000).

Comparing HP-X/HP-Y for Cy3/Cy5 data as 'ratio of ratios'

 a)   (Cy3X/Cy5X1) / (Cy3Y/Cy5Y1)
becomes
 b)   (Cy3X/Cy3Y)

   (Cy3A/Cy5P), (Cy3B/Cy5P), ... , (Cy3N/Cy5P)

  [(Cy3X/Cy5Y) + 1.0/(Cy3Y/Cy5X)]/2

3.1.2 Design philosophy of MAExplorer methodology

The major focus of the MAExplorer is interactive data mining with an emphasis on direct graphical and tabular manipulation of the data. The investigator is able to interact with the system by clicking on spots in the array image, points in graphic plots, cells in spreadsheets, by manipulating threshold sliders or typing in gene names/clone Ids. This level of interaction allows investigators to search for and identify patterns of differences with greater ease than with a more static graphic system since it is easier to test ideas by "grabbing onto the data". For example, "what" is the identity of "this" outlier I am pointing to in a scatter plot; "which" genes are best clustered with "this" gene in this clustergram and are perhaps co-regulated; "which" genes have expression ratios within the range of the histogram bins to that I am pointing?

Direct user manipulation of data, as incorporated in MAExplorer, was defined by (Schneiderman, 1997) who defends the position that the direct manipulation of data in data mining is an extremely effective means to amplify human creativity in understanding patterns. Schneiderman's dogma states "overview first, zoom, and then filter details on demand" and favors the use of "shallow search trees, slide controllers, and information-right screens with tightly coordinated panel view of data", (Beardsly, 1999). MAExplorer also uses many of these direct manipulation principles. It was designed to run on the desktop computers with data residing on the same computer and loaded into its memory for rapid direct manipulation - for both the Web browser and stand-alone versions.

3.1.3 Evolution of MAExplorer from earlier proteomic data mining systems

Part of the Flicker system allows comparison of user 2D gel images with standard images from SWISS-2DPROT for putative identification of unknown spots in the user gels. The user would select a standard 2D gel image from over 20 tissue types, enter their own 2D gel image and align them at spots of interest. They could then switch to a database access mode, click on those spots and generate popup SWISS-2DPROT Web pages for those proteins - similar to Clone reports in MAExplorer. That is accessed at http://www.lecb.ncifcrf.gov/flicker/swissProtIdFlkPair.html.

MAExplorer will have a groupware facility similar to what we have done with our WebGel (http://www.lecb.ncifcrf.gov/webgel/) system described in (Lemkin et al., 1999b). It is a two-dimensional electrophoresis system for sharing data analyses. In WebGel, users may perform a data-mining analysis and leave the state of the their analysis and accompanying notes to share with their collaborators on a login-protected basis.

3.1.4 Concepts used in data mining with MAExplorer

Set operations on gene subsets

User exploration states

User groupware sharing of exploration states with collaborators

We now discuss using these tools for analyzing ones data.

3.2 Steps in data mining, discover, and analysis

In designing a data mining experiment, the first decision to be made is selecting the set of hybridized samples to be compared (steps 1 and 2). This is accomplished by setting the current hybridized sample-X (HP-X) and hybridized sample-Y (HP-Y). In Figure 2.4.4.2 for the scatter plot we selected a single C57B6 pregnancy day 13 and a single Stat5a (-,-) pregnancy day 13 as current HP-X and current HP-Y samples. Changing the normalization changes the view in the scatter plot so that hidden differences may be more apparent (see Figure 2.4.2.3)

The names of the current HP-X and HP-Y samples are displayed at the top of the main window. The current HP-X and HP-Y samples may be changed at any time by clicking on a new sample from a list of samples shown on the left side of the main window or from lists of samples organized by sample population in the Samples menu.

The next decision to be made is selection of the genes to be studied by choosing a subset from the gene class menu list (step 4). Further selection occurs throughout the analysis by clicking on spots in microarray images, points in graphic plots or cells in spreadsheets, by adjusting threshold sliders, or using the text-entry "guesser" to type in gene names, clone IDs, genomic IDs, samples, etc.

The next decision the user must make is to set the intensity data normalization mode (step 3). Normalization of quantitative data is crucial when comparing data between different hybridized microarrays because of spotting, hybridization efficiency, uniformity, and other systematic errors.

Genes of interest may be separated for all of the genes in the database using a cascade of data filters (step 4). Additional filtering options are easily accessible in the (data) Filter menu. Some of the filters require additional parameters. These parameters are set by state scroll bars that pop-up on the screen when data filters requiring them are added to the filter cascade. Changing scroller values causes the data filter to be automatically be reapplied and a new set of genes to be computed.

It is desirable to reduce false-positives found by the data filter by eliminating genes with high quantification variability between duplicate spots on the same sample or spot duplicated in replicate samples. If duplicate genes are available on the array (denoted by Field 1 and Field 2 or F1 and F2 spots), this allows the computation of a coefficient of variation (CV) for the duplicates. This CV may be used in a data filter to reduce potential false-positives. CV is computed as 2|F1-F2|/(F1+F2) using those spot values for each gene, as StdDev_HP/Mean_HP for a set of replicate hybridized samples.

Graphical views of the data give the user additional insights into the data. These include spot intensity and ratio or Zdiff pseudoarray images, scatter plots, histogram plots, expression profile plots, cluster plots showing genes similar to a specified gene, the number of clustered genes for each clone, divisive clusters for K-means clustering, and clustergrams and dendrogams for hierarchical clustering.

Scatter plots are useful for visualizing data from two conditions

Filtering genes by histogram plots of ratios, Zdiffs or intensities

A third method called "K-means" clustering K genes (we call primary nodes) whose expression profiles are most orthogonal to each other. It uses the current gene as the first or "seed" node. It then finds the gene furthest from this and assigns it as node 2. Then the gene furthest from both nodes 1 and 2 is assigned to node 3, etc. This process is repeated until all K nodes are assigned. Then the remaining genes are assigned to the closest node. Having defined the initial cluster centers, it recomputes the centroid of each of the clusters. The centroid can alternatively be computed using a median instead of a mean in which case we would be doing K-median clustering (Bickel, 2001). K genes are then reassigned to the nearest new centroids as the new K-means node instances. Finally, the remaining genes are assigned to the nearest centroid. A scrollable K-means cluster text window report pops up with genes sorted by cluster. Clicking on a gene in either the array image or scatter plot assigns all genes in the cluster to which that gene belongs to the "current cluster". Genes in the current cluster are labeled in the array and scatter plot with a small number of the cluster. In addition, genes in the current cluster are copied to the E.G.L. where they can be used in a report, saved in a named gene set, or used for additional filtering. It also pops up a "N-clusters" scroll bar window to let you dynamically adjust the number of clusters. Changing N will recompute the clusters. When the K-means is recomputed, it uses the current gene as the initial seed gene.

The fourth method is a hierarchical clustering method that generates a clustergram and dendrogram similar to that of Eisen's red-black-green clustergram (Eisen, 1998). This was derived from the clustered correlation map (ClusCor) of Weinstein et al. (Weinstein, 1997). The MAExplorer clustergram and dendrogram are dynamic and may be interrogated and used to set the current gene. This means that it may also position a corresponding ordered list of expression profile plots to the same gene so you may view the data as a plot as well. The dendrogram may be zoomed in to explore a part of the dendrogram in more detail. As with the K-means clustering, a report can be made of the ordered genes.

Gene reports: dynamic spreadsheets for Web access or tab-delimited for exporting data to Excel

Collaborative groupware environment

3.2.1 Definition of expression profile

   ej = (vj1, vj2, vj3, ..., vjN)

   dpq = (1/N SUMj=1:N (vjp - vjp)2 )1/2

 
  spq = 1 - dpq

  djs < T

3.2.2.3 Hierarchical clustering

D can get quite large for clustering a large number of genes N [for N=5000, this is > 50 Mbytes!]

The following is a simplified definition of one way to compute a hierarchical clustering of gene expression profile data.

3.3 Display of gene spot intensity and identification data measurements

     [<field>-<grid name><row#>,<col#>].
e.g.
     [1-A4,3]

If there is only one field in the array, it will appear as field 1. In the above example, [1-A4,3] is field 1 grid A row 4 and column 3. Note that the pseudoarray coordinates are for visualization purposes in MAExplorer and may or may not be the same as the coordinates on the actual array. That depends on how the MAExplorer database was defined in the configuration file described in Appendix C.

When the current gene is defined, it will draw a yellow (green) circle around the spot in the ratio (intensity) pseudoarray image and display other features of the gene in the three-line status area near the top of the main window. If background correction is enabled (the "Use background intensity correction" in the Normalization menu), then spot intensity values will appear as intensity' (with background intensity subtraction) and intensity (without background subtraction).

There are a number of different reporting formats available depending on the array display mode and particular normalization method selected. These include: the pseudoarray image of the intensity of a single sample, the pseudocolor ratio X/Y or Zdiff (X-Y) image (using either HP 'sets' or single samples), or the ratio of Cy3/Cy5 for dual-labeled dyes or F1/F2 for replicate spots for a single sample. In addition, the normalization mode is also displayed in the reporting line. We will present examples of each of these different reporting formats.

You may show the intensity data for a particular spot in the currently displayed pseudoarray image. First select the "Pseudograyscale image" option in the "Show Microarray" submenu in the "Plot menu". If your data has duplicate grids (i.e. fields F1 and F2) then you may look at F1, F2 and mean (F1+F2)/2 data in the reports when you click on a spot. If the "Gang F1-F2 scrolling" switch is disabled in the "View menu", then the intensity value is the intensity data value for the gene at that location. If the "Gang F1-F2 scrolling" switch is enabled, then it reports intensity[F1], intensity[F2], and the F1/F2 ratio. These two formats are shown in the following two examples for a C57B6 pregnancy day 13 samples in the MGAP database:

a) Field F1 spot for a single spot in a single sample with the median intensity selected.

[1-A4,5] intensity=4.5267, (Norm.: median intensity)
CloneID: 1248228, dbEST3': 2279072, GenBankAcc3': AI463183, UniGene: Mm.13859, plate[5,A,5]
GeneName: Mus musculus ribosomal protein L41 mRNA, complete cds

[1-A4,5] intensity[F1]=-0.3067, intensity[F2]=-0.2312, F1-F2=-0.0755, (Norm.: Zscore intensity)
[1-A4,5] intensity[F1]=4.5267, intensity[F2]=6.2408, F1/F2=0.7253, (Norm.: median intensity)
[1-A4,5] intensity[F1]=0.8755, intensity[F2]=1.1457, F1-F2=-0.2701, (Norm.: log median intensity)
[1-A4,5] intensity[F1]=-0.1442, intensity[F2]=-0.0945, F1-F2=-0.0497, (Norm.: Z-score, stdDev, log intensity)
[1-A4,5] intensity[F1]=-0.1533, intensity[F2]=-0.1004, F1-F2=-0.0528, (Norm.: Z-score, mean abs.deviation, log intensity)
[1-A4,5] intensity[F1]=630.9911, intensity[F2]=869.9273, F1/F2=0.7253, (Norm.: calibration DNA intensity)
[1-A4,5] intensity[F1]=1919.9376, intensity[F2]=2646.957, F1/F2=0.7253, (Norm.: scale to max. (65K) intensity)
CloneID: 1248228, dbEST3': 2279072, GenBankAcc3': AI463183, UniGene: Mm.13859, plate[5,A,5]
GeneName: Mus musculus ribosomal protein L41 mRNA, complete cds

c) Ratio data for two samples X and Y in separate hybridized arrays. Ratio data for the field F1 and F2 spot data as well as the mnX/mnY ratio is reported. The median normalization was used in this example.

[1-A4,5] HP-XY: mn(X,Y)=(5.383,6.834) (X/Y)(F1,F2,mean)=(0.651,0.928,0.787), (Norm.: median intensity)
CloneID: 1248228, dbEST3': 2279072, GenBankAcc3': AI463183, UniGene: Mm.13859, plate[5,A,5]
GeneName: Mus musculus ribosomal protein L41 mRNA, complete cds

[1-A4,5] HP-XY: mn(X,Y)=(-0.269,0.151) (X-Y)(F1,F2,mean)=(-0.470,-0.370,-0.420), (Norm.: Zscore intensity)
[1-A4,5] HP-XY: mn(X,Y)=(-0.119,0.051) (X-Y)(F1,F2,mean)=(-0.199,-0.142,-0.170), (Norm.: Z-score, stdDev, log intensity)
[1-A4,5] HP-XY: mn(X,Y)=(1.010,1.224) (X-Y)(F1,F2,mean)=(-0.362,-0.064,-0.213), (Norm.: log median intensity)
CloneID: 1248228, dbEST3': 2279072, GenBankAcc3': AI463183, UniGene: Mm.13859, plate[5,A,5]
GeneName: Mus musculus ribosomal protein L41 mRNA, complete cds

[1-A4,5] intensity[X]=5.3837, intensity[Y]=6.8342, X/Y=0.7877, (Norm.: median intensity)
CloneID: 1248228, dbEST3': 2279072, GenBankAcc3': AI463183, UniGene: Mm.13859, plate[5,A,5]
GeneName: Mus musculus ribosomal protein L41 mRNA, complete cds

[1-A4,5] HP-X 'set' mean intensity=3.295 stdDev=1.482 CV=0.449 n=3, (Norm.: median intensity)
CloneID: 1248228, dbEST3': 2279072, GenBankAcc3': AI463183, UniGene: Mm.13859, plate[5,A,5]
GeneName: Mus musculus ribosomal protein L41 mRNA, complete cds

[1-A4,5] HP-Y 'set' mean intensity=8.180 stdDev=0.986 CV=0.120 n=5, (Norm.: median intensity)
CloneID: 1248228, dbEST3': 2279072, GenBankAcc3': AI463183, UniGene: Mm.13859, plate[5,A,5]
GeneName: Mus musculus ribosomal protein L41 mRNA, complete cds

[1-A4,5] HP-XY 'sets': mn(X,Y)=(3.295,8.180) mnX/mnY=0.402 SD(X,Y)=(1.482,0.986) CV(X,Y)=(0.449,0.120)\
      n(X,Y)=(3,5), (Norm.: median intensity)
CloneID: 1248228, dbEST3': 2279072, GenBankAcc3': AI463183, UniGene: Mm.13859, plate[5,A,5]
GeneName: Mus musculus ribosomal protein L41 mRNA, complete cds

[1-A4,5] HP-XY 'sets': mn(X,Y)=(3.295,8.180) mnX/mnY=0.402 SD(X,Y)=(1.482,0.986) CV(X,Y)=(0.449,0.120) \
        n(X,Y)=(3,5), (Norm.: median intensity)
CloneID: 1248228, dbEST3': 2279072, GenBankAcc3': AI463183, UniGene: Mm.13859, platey[5,A,5]
GeneName: Mus musculus ribosomal protein L41 mRNA, complete cds

[1-A7,20] HP-XY: mn(X,Y)=(3.449,0.853) (X/Y)(F1,F2,mean)=(4.09,4.008,4.041), (Norm.: median intensity)
CloneID: 1382656, dbEST5': 1775754, GenBank 5': AI036495, UniGene: Mm.300,  plate[12,A,8]
GeneName: Carbonic anhydrase 3

Reporting for Cy3 and Cy5 channels for a single hybridized sample

[1-A6,11] Cy5/Cy3=0.3588, Cy5=67.324, Cy3=187.622, (Norm.: median intensity)
CloneID: IMAGE:1054189, 
GeneName: expressed sequence AW213287

Reporting HP-X (Cy3 or Cy5) vs HP-Y (Cy3 or Cy5) data for 2 samples

[1-A5,16] intensX=4.695, intensY=5.923, (X-Y)=-1.2275, (Norm.: log median intensity)
CloneID: IMAGE:963758, 
GeneName: RIKEN cDNA 2410114O14 gene

3.5 Selecting subsets of hybridized sample conditions

3.6 Setting threshold values using the state-scroller sliders

For the intensity and ratio threshold filters, the range interpretation may be inside, or outside the specified range. The ratio range [R1:R2] is between 0.01 and 100.0. The Zdiff range [Z1:Z2] and [CZ1:CZ2] are between -4.0 and +4.0. The intensity threshold range [I1:I2] is set to the dynamic range of the min and max intensity for the current normalization method.

A list of possible threshold sliders is shown in the following table. When a Filter is enabled that requires a slider, it pops up the State Scrollers window that contains one or more slides. When you disable all filters that use these sliders, the popup window will disappear. The corresponding Ratio R1[R2] or Zdiff Z1[Z2] sliders are used if you are using a ratio or Zscore normalization - and will change if the normalization changes while the filter is active.

Some of the sliders are implemented with a non-linear scale so that you have more resolution at the low end (eg. p-Value, Spot CV, Diff HP-XY).

Depending on the set of data Filters selected, there may be multiple sliders present in the State Slider popup window (eg. see Figure 2.4.3).

3.7 Exporting report and plot data

Section 4.1 discusses known bugs, Section 4.2 lists the revision notes for older versions known bugs. If you have experienced bugs with an older version of MAExplorer, you might check the revision notes to see if the bug was fixed and download a new version. Section 4.3 discusses problems in using MAExplorer as an applet with Web browsers. Section 4.4 describes handling fatal "DRYROT" errors.

4.1 Known Bugs in MAExplorer

If you encounter a fatal error that is detected by MAExplorer, it will popup an error reporting window. Please E-mail this data to us so we can try to resolve the problem.

In the mean time, partially implemented commands are disabled to keep you out of trouble :-) ...

You can help us and get MAExplorer to do more of the things you would like to see. Let us know of problems that you encounter as well as suggestions for changes or new methods you would like to see - send us E-mail.

4.1.1 Browser Applet Bugs

4.1.2 Downloading and Installer Bugs

4.1.3 Computation speed and display Bugs

4.1.4 User state and login Status

4.1.5 Data file names Bug

4.1.6 Gene Sets Bugs

4.1.7 Clustering Bugs

4.1.9 Expression profile Bugs

4.1.9 Data conversion problems

4.1.10 Java Plugins Bugs

4.1.1 Browser Applet Bugs

1. When using MAExplorer as a Web browser applet (this does not apply if running as a stand-alone application - and we recommend running it as a stand-alone application not as an applet), there are problems with running MAExplorer on some of the Web browsers on some operating systems. It generally works with Netscape 4.7 or later, on Windows PCs (95/98/NT/2000/XP), and on SUN Unix and with Microsoft Internet Explorer 5.0 on PCs. It may not work as well as an applet on Macintoshes - even with Internet Explorer (although it works fine as a stand-alone application on the Mac) and problems have been reported on SGI systems because of browser problems.

   If you are experiencing Web browser problems using the MAExplorer applet, you might check the discussion of possible solutions.

2. When run as a Web browser applet, you should only click once since multiple clicks may cause some browsers to hang. Note that when you click the first time, it starts downloading the applet. Clicking a second time may cause a second request for the applet to be launched. This interferes with the first request and so neither is started correctly.

3. When the MAExplorer applet is being downloaded, some browsers do not indicate to the user that it is being downloaded and so it may appear that nothing is going on. Some browsers will not indicate that it is downloading an applet. So please be patient. When it starts, the main window will popup and it continues to download data. Wait a while for it to load the applet before giving up. For a 28Kb Internet connection, this could take several minutes. When it is finished, the menu bar will become active and it will display Ready - click on a gene to query database.

1. Because it does a lot of computation, MAExplorer runs best on a fast computer with lots of memory. An optimal system should have at least 128Mb of memory at least 500Mhz CPU speed. For doing hierarchical clustering we recommend 256Mb of memory which will allow you to cluster larger sets of genes without paging the operating system. So you may run into problems with running out of memory on an under-powered computer. To increase the usable memory size to more than 256Mb (assuming your computer has it), then you may need to edit the MAExplorer.lax file that resides where you installed MAExplorer on your computer to increase the heap and stack maximum limit.

2. We have seen instances where the same system will run out of memory with particular Web browsers (again - not a problem when running stand-alone) Netscape 4.7 on a system with a lot of memory. This does not seem to happen with Internet Explorer 5.0. This is because Netscape4.7 on the Windows PC seems to be able to use only about 5.5Mb before it runs out of memory. This is not the case with Internet Explorer 5.0.

3. MAExplorer is difficult to use with a system with a small screen because of the size and number of multiple graphic images and plots. We recommend a screen size of at least 1024x768. As they say these days, more is better :-), generally.

4. For problems with MAExplorer and MacOS, you might also see our MacOS problems FAQ.

5. For problems with MAExplorer and Sun Solaris OS, you might also see our SUNOS problems FAQ.

6. We have on rare occasions seen a data-related problem on MacOS systems. If the data was edited in such a way that the line terminators were Return characters instead of Linefeeds or Return-Linefeeds, the data may become corrupted and MAExplorer has difficulty reading it. The solution is to replace the Return characters with Linefeeds or Return-Linefeeds. One (painful) way this might be done by exporting the data to a system such as a Windows PC; read the data into Excel; save it as an Excel worksheet file format; exit Excel; start Excel again on the new data files; save the data files back into the original MAExplorer directories as tab-delimited text files. MAExplorer should recognize Return as the line terminator. This bug will be resolved in a future release.

1. The full range of GeneClass subsets (Section 2.4.1) is not available yet. However, you may use the gene name guesser to find a set of genes by wildcard into the Edited Gene List (e.g. "*ONCO*" to find all oncogenes or proto-oncogenes, etc). These EGL sets may then be saved as named gene sets and used in the data filter as "Filter by 'User Gene Set'". See the Gene Class ontologies using Gene Set operations (Section 2.4.1) that can achieve a similar effect.

1. The Cvt2Mae data file converter application is currently available for beta-testing on the a related MAExplorer Cvt2Mae Web page. It enables users to convert their data files (academic one-of-a-kind as well as commercial e.g. Affymetrix, Incyte, GenePix, Scanalyze, etc) to the standard MAExplorer files as described in Appendix C and D.

2. If you are not able to convert your data using Cvt2Mae (which is currently in early Beta testing - see the Cvt2Mae Web page for current status), you can always manually convert the data using the examples in Appendix C.

3. We have seen some problems running the converter with some types of data on MacOS 8-9. The temporary solution is to convert your data using another system such as Windows until we fix the problem.

4. If you are using the NCI-CIT mAdb system to package data for MAExplorer, you must select all samples from multiple projects that you want to compare with MAExplorer while packing the data in mAdb. (See PDF document "How to download mAdb data For MAExplorer") Currently, you can not easily merge the data after you have downloaded the files to your computer. The solution is to use mAdb to prepackage all data you want to analyze in a single .zip file. You can not currently merge data from separately unzipped files (without manually editing the SamplesDB.txt file). The way you package data using mAdb is to: 1) select ALL of the projects that contain samples that you want (on the mAdb "Top Level Analysis Selection" web page); 2) further select the exact subset of samples that you want in the "Array Selection" table on the "mAdb: Database Retrieval" web page). NOTE: currently, you can only select arrays using the same GAL file so they have compatible geometries. This is on our list of things to change so you can mix and match data from different chips.

4.2 Revision notes

Note: An archive of some of the stable older releases is available on the NCI/LECB Web site for a limited period.

• Version 0.96.34.01: (11-24-2003) Fixed problem introduced in the previous release where you could not increase the memory size invoked by the (Edit | Preferences | Resize MAExplorer memory limits for the next time it is run) command. To fix this bug you must reinstall MAExplorer using the full download using the Java installers. Also added ability to visualize the CV (coefficient of variation of a set of replicate samples) using the new (Analysis | Plots | Show microarray | Pseudocolor HP-EP 'list' CV (Coefficient of Variation) [RB]) command. It computes the CV for the HP list. So use the (Samples | Choose HP-X, HP-Y and HP-E samples) command to define the HP-EP list of samples you want to investigate.

• Version 0.96.33.07: (08-15-2003) The code implementing the R eval paragigm has been refactord so that it is clearer and will be easier to use for other projects. See the RtestPlugin documentation Section 6 for details.

• Version 0.96.33.04: (07-31-2003) Fixed minor bugs in MAExplorer and Reval data export for RLO methods.

• Version 0.96.33.03: (07-23-2003) Fixed minor bugs in current OCL state as well as adding support for future MAERlibr.zip updates from the Web.

• Version 0.96.33.01: (07-09-2003) Fixed bugs where the XY scatterplot and other functions may not work with some data where there were empty genes in the database. Changed the underlying control mechanism to switch between the 4 clustering methods so it is more robust and easier to use.

• Version 0.96.32.10: (06-30-2003) added loessPlot() R method to MAERlibr R library. This is used by RtestPlugin for generated code.

• Version 0.96.32.09: (06-26-2003) added new command (Plugin menu | Save RLO reports in time-stamped Report/ folder [CB]). put files generated by R from successive executions of the same RLO into separate sub-folders in the Report/ folder with names "RLOname-YYMMDD-HHMMSS/" to peep the data separate. This is useful when you want to compare results from the same RLO method but with different MAExplorer preprocessing. Also added new command (Analysis menu | Filter | Filter by genes with non-zero intensity) which can be used to omit genes with zero intensity values in any samples of the HP-EP set of samples. This is useful when exporting data to RLOs that may have problems by dividing by 0.0 etc.

• Version 0.96.32.08: (06-17-2003) fixed bug in RtestPlugin for Version 0.96.32.07 where it did not always use the intput file mode correctly. Simplied the "Add Template" usage to "Generate Template for 1st input file". Finished inplementing the (Files | Update RLO methods from Web server) so that it updates the MAERlibr package library file (lib/MAERlibr/R/MAERlibr) as well as the .R and .rlo files for the RLO methods.

• Version 0.96.32.07: (06-15-2003) fixed RtestPlugin editor bug that did not always setup parameters for New RLO correctly. There is still a but that may require you to set the input files mode to the corresponding Demo BEFORE doing "Add Template". This will be corrected shortly.

• Version 0.96.32.06: (06-13-2003) fixed RtestPlugin editor bug that did not always save the Routput files types. Also fixed bug in demo R script Test.R.

• Version 0.96.32.05: (06-12-2003) fixed code generators and changed all R methods in the MAExplorer MAERlibr to have a "mae." prefix to minimize conflict with other packages. Added alpha-level support for XY overlay plots in MJAplot class of the API.

• Version 0.96.32.04: (06-11-2003) fixed code generators on Demos 1-15. Demos 16-17 now broken with VSTACK error. Major code generators cleanup in RtestPlugin. Demo 15 (with use Filter Action) works but does not toggle the (Analysis | Filter | Filter by user gene set) menu checkbox. Also added code to R2MAE actions to better support reimporting gene sets back into MAExplorer.

• Version 0.96.32.03: (06-08-2003) Moved the MAExplorer-R interface R library code from the .R files to an full R library that now is downloaded when you do the full download/install of MAExplorer. We will be adding an update for this MAERlibr library. The Web update is not available currently available.

• Version 0.96.32.01: (05-29-2003) Added (Files | Update RLO methods from maexplorer.sourceforge.net) command to update RLO methods from the server. RLO methods consist of pairs of R scripts and RLO files. Note: R must be installed on your computer to use these methods. When MAExplorer is restarted these new RLO methods will be available in the (Plugins | RLO methods) submenu.

• Version 0.96.31.23: (05-27-2003) Fixed bug in reading RLOs if they were manually edited with a Windows text editor that replace line feeds with carriage returns.

• Version 0.96.31.22: (05-22-2003) Minor changes in RtestPlugin and R evaluation support code.

• Version 0.96.31.21: (05-18-2003) Reorganized RtestPlugin code and updated some of the code generators.

• Version 0.96.31.20: (05-14-2003) Fixed bug where MAExplorer could crash if R was not installed. This was also in the RtestPlugin. Now, it detects that it is not there if you cancel the prompt for specifying R is when it can't find it. This is noted in a special .RbasePath file which contains "**NO_R_INSTALLED**". This is then used to prevent you trying to access any R commands or menus (i.e. keep you out of trouble).

• Version 0.96.31.19: (05-14-2003) Fixed minor bugs in RtestPlugin and support for horizontal and vertical stacked data.

• Version 0.96.31.18: (05-08-2003) Fixed minor bugs in RtestPlugin and support.

• Version 0.96.31.17: (05-06-2003) Fixed prompt in "Update Plugins".

• Version 0.96.31.16: (05-05-2003) Changed the "Update MAEPlugins from maexplorer.sourceforge.net" so that it will download ALL of the plugins if you answer YES rather than one at a time. More cleanup for RtestPlugin Template paradigm.

• Version 0.96.31.15: (05-05-2003) In RtestPlugin, simplified R code to use new multi-line header data file paradigm and started to add data model Template generation code. OCL demo does not work with new paradigm.

• Version 0.96.31.14: (04-30-2003) Fixed problem with R script evaluation if the sample names had spaces in the name. Quoting these fixed the problem.

• Version 0.96.31.12: (04-28-2003) Fixed fatal error problem introduced in 0.96.31.12 if start MAExplorer with no database. This version fixes bugs in some of the RtestPlugin Demo R code generators and data files.

• Version 0.96.31.11: (04-25-2003) Fixed quote problem with with executing R in some cases for RtestPlugin. Fixed Prev/Next problem.

• Version 0.96.31.10: (04-23-2003) Added "Save ALL Demo RLOs" to RtestPlugin to generated demo RLOs for (Plugins | RLO methods) menu next time MAExplorer is restarted. The demo R Code generators are not validated yet.

• Version 0.96.31.9: (04-22-2003) Fixed RtestPlugin problem with sample set names when generate RLOs for XY sets and OCL condition sample sets. Changed the export data formats so easier to use in R. The demo R Code generators are not validated yet.

• Version 0.96.31.8: (04-21-2003) Fixed problem with RtestPlugin that sometimes showed up when installed in "C:/Program Files/" on a Windows system (needed to quote the path). It now works on windows. Note: RtestPlugin Demos 10-13 and Demo 16 are not working correctly. The demo R Code generators are not validated yet.

• Version 0.96.31.7: (04-18-2003) Gene Set actions now work for RLOs. The demo R Code generators are not validated yet.

• Version 0.96.31.5: (04-13-2003) Extended the RLO definition and support code. Extended RtestPlugin to access this. The demo R Code generators are not validated yet.

• Version 0.96.31.4: (04-11-2003) Added "RLO methods to Plugins menu. This uses RLO files generated by the RtestPlugin. Additional bug fixes and better error reporting of R evaluation errors. The demo R Code generators are not validated yet.

• Version 0.96.31.2: (04-07-2003) Extended RtestPlugin and corresponding changes in MJA API. The demo R Code generators are not validated yet.

• Version 0.96.31.1: (04-03-2003) Cleaned up gene filter processing model.

• Version 0.96.30.10: (03-29-2003) Better support for RtestPlugin.

• Version 0.96.30.8: (03-28-2003) Modified the MJAReval, MJAgeneList, MJAfilter API class methods to support the RtestPlugin. RtestPlugin now supports "New RLO" at alpha level.

• Version 0.96.30.6: (03-25-2003) The RtestPlugin editor is working (alpha level) for demonstration purposes-only at this time.

• Version 0.96.30.5: (03-18-2003) The RtestPlugin is alpha level. Modified the MJAReval class methods to support this. This is still alpha-level, and may change.

• Version 0.96.30.4: (03-17-2003) Added new functionality to evaluate MAExplorer data using the R system. There is a new MJAReval to access methods to do this. Using MJAReval, one can write MAEPlugins to process MAExplorer data in R and to bring back the results into MAExplorer. This is still alpha-level, and may change.

• Version 0.96.30.3: (03-04-2003) Fixed radio-button usage of NormalizationPlugin so that only one normalization (either built-in or plugin) is active at a time. The MJA API was enhanced to add this functionality.

• Version 0.96.30.2: (03-02-2003) Major ehnancements of the NormalizationPlugin paradigm in the MJA library to support a wide variety of normalization methods - both global and local. This includes support code throughout MAExplorer to support it and to make it easier to debug. Although it works, the NormalizationPlugin is still Alpha-level and will be documented fully when it enters Beta-level code.

• Version 0.96.29.3: (02-25-2003) Rebuilt installers (Windows, Linux, Solaris) with Java Virtual Machine (JDK) 1.4.1_01 that has improved performance. You must do a download and reinstall -- not just an update from (File menu | Update MAExplorer from maexplorer.sourceforge.net).

• Version 0.96.29.3: (02-23-2003) Added more support for FilterPlugins. The ExampleXYdataFilterPlugin example now works. It is meant to be a model for more sophisticated plugins. [TODO] it needs parameters GUI popdown.

• Version 0.96.29.2: (02-21-2003) Fixed many problems with MJA classes where some methods which should have been public were not so MAEPlugins could not access them. This has been corrected. The FtestNconditionsFilterPlugin is now working.

• Version 0.96.29.1: (02-19-2003) Added popup alert message window for bettering informing users of conditions that prevent them from doing the operation they requrested. They must press the Close button to pop-down the message, although they can do a SaveAs to a file to save the message. For complex problems, some of the messages may suggest what they need to do to correct the problem.

• Version 0.96.28.3: (02-18-2003) Added 2 new filters: "Filter by HP-X,HP-Y 'sets' t-Test [p-Value] slider [RB]" and "Filter by current Ordered Cond. List (OCL) F-Test [p-Value] slider [RB]". The p-Value related Filter tests were changed to radio buttons [RB] so that only one would be operative at a time. If another p-Value dependent test is on, it will be shut off it you select a different p-Value dependent test. Added new support methods in the MJA API in classes MJAsampleList, MJAstatistics, MJAfilter, and MJAmath. Implemented the "List OCL" and "List All" OCLs in the OCL chooser. Improved the GUI legends in the Condition Chooser and the Ordered Condition Chooser.

• Version 0.96.28.1: (02-15-2003) fixed default R2 and CR2. Modified State Scrollers paradigm so that it uses slider usage counters (instead of flags) so that MAEPlugins may share sliders (e.g., p-Value, cluster distance, etc).

• Version 0.96.27.1: (02-12-2003) added named Ordered Condition List (OCL) editor as a builting method.

• Version 0.96.26.1: (02-08-2003) added methods to MJAsampleList to get either normalized Filtered data or normalized complete set of genes data. Fixed bug in CompositeDatabase.

• Version 0.96.25.5: (01-28-2003) Fixed bug in "Remove Parameter" in the condition chooser. Added new methods MJAcondition, added method in MJAsampleList to get all raw sample data.

• Version 0.96.25.4: (01-13-2003) Fixed bug in "Remove Parameter" in the condition chooser.

• Version 0.96.25.3: (01-02-2003) Fixed bug in "Remove Condition" where it previously deleted the wrong condition list.

• Version 0.96.25.2: (12-14-2002) Added "Remove Condition" to (Sample menu | Choose named condition lists of samples) as well as fixed several bugs related to the condition chooser. Also added the same command to the (Edit menu | Sets of conditions | Choose named condition lists of samples) menu since it is a condition editing function.

• Version 0.96.25.1: (12-11-2002) Made some of the tests for names case-independent for smoother operation.

• Version 0.96.25: (12-06-2002) Added new a new command (Sample menu | Choose named condition lists of samples) lets you define or edit new named lists of hybridized samples.

• Version 0.96.24.12: (12-04-2002) Added support for RefSeqID and alternate spellings of other genomic identifiers. Modified MAJcondition class to support ConditionChooser.

• Version 0.96.24.11: (12-02-2002) Added support for RefSeqID and alternate spellings of other genomic identifiers. Modified MAJcondition class to support ConditionChooser.

• Version 0.96.24.10: (11-30-2002) Updated the default URLs used for GenBank, UniGene, LocusLink, and added OMIM ID support.

• Cvt2Mae Version 0.73 (11-26-2002) Added ability specify multiple chips/sample (e.g. Affy U133A and U133B) for the MAS5.0 data. It will merge the data into one sample during the conversion.

• Version 0.96.24.8: (11-20-2002) The ConditionChooserPlugin is now working (alpha) and is included in the distribution. Additional changes were made to MAExplorer to support it.

• Version 0.96.24.7: (11-18-2002) Fixed Scrollbar initialization problem for the non-linear scrollers (p-Values, CV value, etc.)

• Version 0.96.24.6: (11-15-2002) Fixed Q&A error in the command "Resize MAExplorer memory limits for the next time it is run" is in the (Edit | Preferences) menu. Added or debugged methods to MJAproperty and extended MJAcondition MJA API.

• Version 0.96.24.5: (11-07-2002) Add command to resize the memory limits of MAExplorer. The command "Resize MAExplorer memory limits for the next time it is run" is in the (Edit | Preferences) menu. After the command is run, you must exit and restart MAExplorer for the new memory limits to take affect. It changes the memory limits in the MAExplorer.lax startup file.

• Cvt2Mae Version 0.72.4 (11-07-2002) Fixed bugs in "Define Gipo Fields" where it would be be available under certain conditions. Also fixed another bug introduced with the addition of the MAS5.0 data.

• Cvt2Mae Version 0.72.3 (11-05-2002) Increase buffer sizes to fix problem for very large input files with large gene descriptions.

• Version 0.96.24.4: (11-03-2002) Fixed bug in non-linear sliders in State Scrollers. This had affected the values for the p-value and CV value. Extended functionality of popup State Scrollers sliders window. You can now toggle between all sliders, even those not active, and only those that are active.

• Cvt2Mae Version 0.72.2 (11-02-2002) Fixed bugs in Affymetrix MAS5.0 where the "Define Quant Fields" were not being mapped automatically and separate files were not converting correctly.

• Version 0.96.24.2: (10-24-2002) Added (File | Update Plugins) operation. It will prompt you on whether you want to update each default plugin one by one.

• Cvt2Mae Version 0.71.2 (10-23-2002) Moved the name of the available Cvt2Mae.jar version number to the popup dialog query when doing the "Update Cvt2Mae" operation.

• Version 0.96.24: (10-22-2002) Extended the number of grid groups from 52 [A-Z,a-z] to 152 [A-Z,a-z,1-100]. Moved the name of the available MAExplorer.jar version number to the popup dialog query when doing a (File | Update MAExplorer) operation.

• Version 0.96.23: (10-21-2002) Extended the MJAgene.java API class so MJAPlugins can set gene annotations and properties from the plugin using external data.

• Version 0.96.22: (10-18-2002) Extended the (File | Update MAExplorer) command so that it reads the version number of the MAExplorer.jar file on the maexplorer.sourceforge.net server and displays it prior to requesting the user to finish requesting the update. This allows users to see if they want to do the update.

• Cvt2Mae Version 0.71.1 (10-16-2002) Added a "Update Cvt2Mae" button. This will (1) backup the current Cvt2Mae.jar file as Cvt2Mae.jar.bkup; (2) copy the latest Cvt2Mae.jar file from the maexplorer.sourceforge.net Web site and replace your Cvt2Mae.jar file in your installation directory. Then when you restart MAExplorer, it will use the new version of the program. The much more time consuming alternative is to do an entire download and reinstallation from the Web site.

• Version 0.96.21: (10-15-2002) fixed comments for (File | Update MAExplorer) to give the actual server it will use. Fixed default PseudoArray Image to use when have single Cy3/Cy5 sample to use (Red-Yellow-Green) display. Cvt2Mae Version 0.70.7 (10-15-2002) has a major re-indentation of the Java source code to make it easier to read using the convention describe in the previous revision.

• Version 0.96.20: (10-14-2002) We added a "Define GEO Platform ID" command for arrays that were submitted to NCBI's GEO (Gene Expression Omnibus) for use with future MAEPlugins. You can now use the new "Update MAExplorer" command in the File menu. This will (1) backup the current MAExplorer.jar file as MAExplorer.jar.bkup; (2) copy the latest MAExplorer.jar file from the maexplorer.sourceforge.net Web site and replace your MAExplorer.jar file in your installation directory. Then when you restart MAExplorer, it will use the new version of the program. The much more time consuming alternative is to do an entire download and reinstallation from the Web site. This version also has a major re-indentation of the Java source code to make it easier to read. It uses the automatic java indentation engine in Fote 4.0 (2 spaces/tab, convert tabs/space, space before '{', no space before '(').

• Cvt2Mae Version 0.70.6 (10-11-2002) Fixed bug in Configuration file generator.

• Version 0.96.18: (10-09-2002) Changed popup HP chooser initial window sizing so it works more consistently across platforms.

• Version 0.96.17: (10-04-2002) Fixed bug with DetValue slider (the test was reversed). Also increased the dynamic range and precision of value indicated. The default is to leave the Detection Value filter off. Changed the default PseudoArray Image if there is one Cy3/Cy5 sample from the intensity display to the sum of cy5 (red) and Cy3(green) channels.

• Cvt2Mae Version 0.70.5 (10-03-2002) Fixed bug in Affy MAS4 converter when selected "Generate genomic IDs from Descriptiopn". Also fixed bug in Quant Field assignments so it now shows the "DetValue" field when it was selected by the DetValue checkbox in the Genomic Identifiers Wizard.

• Version 0.96.16: (9-30-2002) Fixed bug with positive data filter whereby it defaulted to being on. This was changed so that you need to explicitly enable it in the Filter menu.

• Cvt2Mae Version 0.70.4 (09-27-2002) Added GEO Platform ID support.

• Cvt2Mae Version 0.70.3 (09-23-2002) fixing the ArrayLayout Affy MAS5 tmp file converter. It now appears to generate the temp file correctly and read it correctly. Need to verify the code generators for the GIPO.txt file. There was a problem with embedded double quotes and the MAS5 ALO was not correct. Also had to add the additional fields to the FieldMap when the tmp file is extended.

• Version 0.96.15: (9-19-2002) Added a new "Filter by Spot Detection Value" data filter. This can be used with the new Affymetrix MAS5.0 "Detection p-value" mapped through the "DetValue" or "CorrCoef" field in the .quant file. The next version of Cvt2Mae after 0.70.2 will output the "DetValue" field.

• Cvt2Mae Version 0.70.2: (09-18-2002) Added automatic delete of old default Affymetrix*.alo files when Cvt2Mae starts up. This was required since we changed the names to reflect the MAS-4.0 and MAS-5.0 versions of the analysis software. Without this, you would have multiple names for the same files in the Array Layout chip list.

• Cvt2Mae Version 0.70.1: (09-17-2002) Fixed bugs introduced in 0.70. Added remap code to better handle MAS5.0 data.

• Cvt2Mae Version 0.70: (09-15-2002) Added Affymetrix MAS5 array layout so it will map (Signal, Detection, Detection p-value) to (RawIntensity, QualCheck, CorrCoef) Quant data fields. Fixed bug in "SwissProtID" data parser.

• Version 0.96.14: (9-14-2002) Fixed incorrect base URL in "Help" pull-down menu. It had pointed to the old MAExplorer Web site not the the new http://maexplorer.sourceforge.net/MAExplorer/. The same documents were always visible from the Web site, but this makes it more convenient.

• Cvt2Mae Version 0.68: (08-30-2002) Fixed bugs in Edit Layout where it was not always saving the changes you make to the state. Pressing Cancel in the Edit Layout restores the initial state before starting the edit. Pressing Back saves the state of the current panel you are working - previously only Next or Finish did that.

• Version 0.96.12: Fixed bug introduced in version 0.96.11 such that it would pop up the dialog query when you did a "Open Disk DB" command.

• Version 0.96.11: Extended MJAeval methods so can more easily invoke a new .mae startup file via a client-server MAEPlugin as well as the invokeCommand method. Modified internal methods and structures to support this.

• Cvt2Mae Version 0.67: (08-15-2002) Fixed bug where the generated pseudoarray image was overiding the user specified (grid,row,column) geometry.

• Version 0.96.10: Fixed bug in pseudoarray image whereby the current gene icon (yellow circle) sometimes was not reset properly for a database with X,Y coordinate data if the current gene was selected and you switched the current sample.

• Version 0.96.08: Fixed bug in pseudoarray image whereby the current gene icon (yellow circle) sometimes was not reset properly if you toggled the mouseover checkbox, selected a different sample using the sample list in the pseudoarray image, or scrolled the pseudoarray.

• Version 0.96.07: Fixed bug in font size updates for the pseudoarray image and scatter plot. Added methods to MJAproperty to get font size and font familiy.

• Version 0.96.06: Modified data filter property bits for Gene, MJAgene, and Filter to make them more consistent when writing MAEPlugins.

• Version 0.96.05: Fixed bug in data filter that caused problems with FilterPlugin's.

• Version 0.96.04: Fixed bug when run database with no startup database, it does not refresh spot lists in the PseudoArray image.

• Version 0.96.02: Fixed bug when run database with no startup database and no default MGAP demo database. It now lets you search the disk to find a .mae startup file instead of aborting with a Dryrot error. Fixed bug with chooser where it sometimes fails to let you access the last element of a list. The NCI/LECB site and the SourceForge site now both have the same full installers with JDKs (except for AIX and HPUX) and with the MGAP database.
  A Cvt2Mae Version 0.66: bug has been fixed that makes it easier to automatically find the first row of spot data when scanning the users data input data file. Previously, the user might have to manually enter that starting data arow number in the Edit Layout wizard.

• Version 0.96.01: Initial release on . Note: The NCI/LECB site will have the same installers., but with JVMs if you want to download a version with the JVM for your computer. The SourceForce does not have the JVMs because of space limitations.

  ---

• Version 0.95.20: Inserted Open Source License into all code.

• Version 0.95.19: Changed default for (File | Save file DB) and (File | SaveAs file DB) so that it saves all samples in the new .mae startup file - not just the samples in HP-X 'set', HP-Y 'set or in HP-E 'list'. Made the three status lines in the main window non-editable.

• Version 0.95.18: Maintenance release. The Open Java API has been changed in the mjaGeneList, mjaGene, mjaEval classes.

• Version 0.95.17: Maintenance release. The Open Java API has been updated and now includes working updateCurGene(), updateFilter(), updateSlider(), updateLabel(), and close() methods as well as new methods in the MJAgeneList API class. The stable release is now the version with the MaeJavaAPI active. The previous stable release 0.95.04 did not have the API.

• Version 0.95.16: Maintenance release.

• Version 0.95.15: Maintenance release to fixed FilterPlugins so the FilterTestPlugin example is working. Fixed bug recently introduced that prevented selecting spot in pseudoarray image.

• Version 0.95.14: Fixed bug recently introduced that prevented menu commands from working correctly.

• Version 0.95.12: Maintenance release. Changes to help support MAEPlugins and additional functions in the Open Java API. Fixed Quant input data parser bug so it now handles strictly lower-case letters for (Grid,Row,Column) data.

• Version 0.95.11: Maintenance release. Changes to help support MAEPlugins and additional functions in the Open Java API.

• Version 0.95.10: Maintenance release changes to help support MAEPlugins and additional functions in the Open Java API. Menu Plugin command is enabled to load MAEPlugins built using the Open Java API (under construction).

• Version 0.95.08: Maintenance release to help support MAEPlugins and additional functions in the Open Java API. Menu Plugin command is enabled to load MAEPlugins built using the Open Java API (under construction). Added Plugins/ folder to distribution that is installed where you install MAExplorer. It contains several alpha-level JAR file plugins that work with the (Plugins | Load Plugin) command. We will be releasing the source code soon as the Open Java API alpha release in put on the MAEPlugins Web site.

• Version 0.95.05: Maintenance release. Changes to help support MAEPlugins and additional functions in the Open Java API. Menu Plugin command is enabled to load MAEPlugins built using the Open Java API (under construction). Added (Plugins | Test Plugin Code) to invoke code for a plugin a debugging context (maintenance).

• Version 0.95.04: Same as Version 0.95.03 optimized. It was compiled with the optimizing Java compile without debugging symbol tables. All Plugin support codes was removed. This results in a MAExplorer.jar file of 449Kb vs 599Kb for. Since the MGAP Web site Applet version can not use Plugins, this version is optimized for Applets and is to preferred over the Beta 0.95.03 version for now.

• Version 0.95.03: Maintenance release. Changes to help support MAEPlugins. Menu Plugin command is enabled, but will only load MAEPlugins built using the Open Java API (under construction).

• Version 0.95.01: Increased the number of spots/array that may be used from 16K to unlimited. Maintenance release. Changes to help support MAEPlugins.

• Version 0.94.15: Fixed bug where occasionally, it would not load additional samples from the (Samples | Set Samples from Lists | ...) commands.

• Version 0.94.14: Maintenance release. Changes to help support MAEPlugins. This release fixes a menu-related bug that occured if you used (File |Open File DB) to switch to a different database or to start the initial database rather than starting directly from the .mae startup file. The problem was that some of the sub menus generated might not correspond to the type of database. For example, if the previous database was an intensity database with duplicates F1 and F2, switching to a Cy3/Cy5 database might have some of the menus still appear as (F1/F2).

• Version 0.94.12: Maintenance release. Changed placement of magenta circle marker so it is more visible in ClusterGram when selecting genes in the dendrogram/ClusterGram for the EGL by clicking on a gene with the Control key pressed.

• Version 0.94.11: Fixed fatal error introduced ni 0.94.10 that occurs if you start the database with no data. Version 0.94.10 did work if you started it with a particular .mae startup file.

• Version 0.94.10: Maintenance release. The "Filter by positive Quant data" toggle checkbox in the (Analysis | Filter menu) is enabled for all databases - not just those requesting it. If the database has 2 channels (F1, F2) or (Cy3,Cy5) each channel is checked. If the background correction is enabled, the background corrected values are tested to see if any of them are negative. Changed the labels in the scatter plot so that instead of displaying the current gene values as intensX= and intensY= or intens'X= and intens'Y= , it now displays it as X= and Y= or X'= and Y'=. The ' indicates that background correction is active. A new (Analysis | Plot | Show Microarray | "Scale pseudoarray image by 1/100 to zoom low-range values) command has been added to rescale intensity and (Cy3+Cy5) and (HP-X + HY-Y) (Red-Yellow-Green) plots to so low values are easier to visualize.

• Version 0.94.09: Maintenance and Reference Manual update. Added copy of "Show 'Edited Gene List'" to Edit EGL submenu so easier to find.

• Version 0.94.08: Changed button layouts on popup plots so it is easier to resize them to remove the right part of the plot if you want to conserve screen space.

• Version 0.94.07: Fixed bug recently introduced in the histogram plots which prevent them from working. Fixed another histogram plot bug where the frequency count for "><" thresholding was reported incorrectly.

• Version 0.94.06: Added the saving of the current values of all of the State Threshold scrollers when the (Edit menu | Preferences | Adjust all Filter threshold scrollers) in the message logging area when the State Thresholds popup window is closed. The Cvt2Mae (Version 0.60) has been extended to now handle Scanalyze data files as well as supporing a separate GIPO input data file.

• Version 0.94.05: Added logging of MAExplorer messages to buttons in popup plots and reports. Fixed bug where old fixed gene sets for empty wells were not always restored correctly fro the previous state.

• Version 0.94.04: Added logging of MAExplorer messages and command history in popup windows in the View Menu as "Show log messages" that occur during a session and "Show log of command history" respectively. The windows may be saved in log files. Added additional error detection of bad (field,grid,row,column) data when reading the GIPO and Quant files.

• Version 0.94.03: Maintenance release for improving MAEPlugin support.

• Version 0.94.02: Minor bug fixes related to the renaming of HybProbe to Samples menus. Also, extensive update of reference manual in both text and figures to showing many of the renamed menus so it is synchronized with the program.

• Version 0.94.01: Major version release.

  Renamed all previous references in the program to "hybridization probe" or "hybridization sample probe" to the new term "hybridized sample" for clarity. Changed many "HybProbe Menu" to "Samples Menu" and also many menu selections as well as plot and report labels to reflect this change. We are in the process of updating the manual computer screen figures and PDF slide presentations so that "hybridization sample probe" is shown as "hybridized sample". Also fixed other minor problems including fixing the inverted color scale for the "Pseudocolor Red(Cy5)-Yellow-Green(Cy3) Cy5+Cy3 ratio or Zdiff" command.

• Version 0.93.03: Minor reorganization of (Plots | Clusters) submenus. This is the last release using the "HybProbe" menu.

• Version 0.93.02: Beta-test of JDK-1.3 MAExplorer.jar file and new InstallAnywhere 4.5 installer. This adds JDK-1.3 installers for MacOS-X, IBM AIX, HP-UX, Linux. However, MacOS 8.1-9.x use the MRJ-2.2.5 JDK1.1.8 library. It also fixes a problem recently found running MAExplorer on some Windows-NT systems.

• Version 0.93.01: Major version release.

  Moved "Cluster Plots" submenu of the "Plots" submenu up one level in the "Analysis" menu.

• Version 0.92.24: Maintenance and added more Plugins support. Top level Plugins menu appears, but is disabled in production version.

• Version 0.92.23: Maintenance and changed names of some commands and some messages. Major edit of Reference Manual including expanded table of contents.

• Version 0.92.22: Last Stable Release.

  Major version release. Optimized colors in grayscale display for "Pseudocolor (HP-X,HP-Y) 'sets' p-value". Fixed error and optimized t-test computation.

• Version 0.92.21: Changed grayscale display for "Pseudocolor (HP-X,HP-Y) 'sets' p-value" to display it as a color spectrum. Changed data displayed so p-Value is in the front of the list when you click on a spot etc. Added tests to disallow this display unless the data supports it.

• Version 0.92.20: Beta testing new pseudoarray image display "Pseudocolor (HP-X,HP-Y) 'sets' p-value" that displays a gray value proportional to the p-Value in a t-Test of the HP-X 'set' vs HP-Y 'set'.

• Version 0.92.18: Improved EP plot labels so it no longer prints the "MID:" prefix. Improved spot features list so it no longer prints the "plate[null,null,null]" if the plate coordinates are not present in the GIPO file.

• Version 0.92.17: Fixed a bug which incorrectly computed the max and minimum ratio Cy3/Cy5 ratio values if the data allows negative intensity on these channels after background subtraction. This is required to support the GenePix array data format converter in the new version of Cvt2Mae (11-24-2001).

• Version 0.92.15: A new (Edit | Preferences) command Adjust all Filter threshold scrollers - popup the state scroller window with all of the thresholds. This is useful when you want to adjust thresholds before you enable data Filtering or clustering. A bug was fixed where the state of the previous clustering method was not always cleared if you aborted clustering by clicking on the delete window button (eg. in Windows or the Mac). Click on a row with the Control (Shift) key pressed in the ClusterGram for hierarchical clustering will add(remove) that gene to(from) the Edited Gene List. If the (View | Show 'Edited Gene List') is enabled, then it will draw a magenta '*' before the gene name to indicate that you have selected it. This lets you select a particular subset of the genes from the hierarchical cluster. The Cvt2Mae data converter now has an alpha-release to let you convert non-standard data using the <User-defined> data option.

• Version 0.92.14: The name of sample names and sample file names IDs has been made more flexible. If the "Database_File" is longer than 32 characters, there may be problems reading that file with MacOS-8/9. Therefore, you can use the "DatabaseFileID" to specify the "Database_File" (in which case they would be the same). Then, the labels used for the samples are the "Sample_ID" fields. This is upwards compatible with the previous method which used the "Database_File" for both the file name and the label name. The Reports of Highest (Lowest) Cy3/Cy5 data for a single sample were not easily accessible and this has been fixed. Alternate names for automatic detection of ESTs, ESTs similar to named genes were expanded and also includes the "EmptyWell" gene set (i.e. spots with no genes) (see Automatic Gene Class naming based on Gene Name in Appendix C Table C.4.1). A clustering option was added, Use median instead of mean for K-means clustering, of mean when computing K-means clustering (Bickel, 2001)).

• Version 0.92.13: Maintenance release. Changed per-sample Good Spot filter algorithm and the CV filter algorithm. Added options to filter by either "HP-X 'set'" or "HP-Y 'set'". Algorithm now uses filters "HP-X or HP-Y" and "HP-X or HP-Y 'sets'" rather than the previous "HP-X and HP-Y" and "HP-X and HP-Y 'sets'". Adding some of the missing state that was not being saved during a "SaveAs DB" operation when creating a checking .mae startup file.

• Version 0.92.12: Maintenance release for Plugins.

• Version 0.92.11: Maintainence release. Increased width of scrollable sample names selection windows so it is easier to see the details on long sample names.

• Version 0.92.10: Maintenance release. Due to a problem that some users are experiencing with the Java Installer, we are reverting to the previous version 4.01 (that does not support Mac-x). We will resolve these problems as soon as possible. Also fixed a problem with the Pseudocolor Red(Cy5)-Yellow-Green(Cy3) Cy3/Cy5 ratio or Zdiff which is now resolved.

• Version 0.92.09: Maintenance release. Distribution of Java application with InstallAnywhere 4.5 (previous was version 4.0.1). This now supports MacOS-X, other Unix systems and may fix some other installation problems.

• Version 0.92.08: Fixed bug in Edit use (Cy5/Cy3) else (Cy3/Cy5) for each HP command in HybProbe menu that changes were not saved in the state. Fixed bug introduced in HP Cy3 vs Cy5 intensity scatter plot.

• Version 0.92.07: Added display of Cy3 and Cy5 channel intensity when click on gene in Pseudocolor Red-Yellow-Green Cy3/Cy5 ratio or Zdiff mode. Also added additional support for plugin installation from Configuration DB and .mae startup files.

• Version 0.92.06: The Pseudocolor Red-Yellow-Green Cy3/Cy5 ratio or Zdiff - display the Cy3 and Cy5 microarrays as a pseudocolor image as the sum of Cy3 (Red) and Cy5 (Green).

• Version 0.92.04: Internal changes. Also fixed bug in reading Affymetrix data converted with Cvt2Mae in the case where there is no genomic identifier and the "Location" is used as the genomic identifier.

• Version 0.91.03: Made configuration database "swapRowsColumns" default to FALSE. This should probably never need to be true except for the initial MGAP database. If this flag was not set to false in previous configuration DB files, it would swap rows and columns in the pseudoarray image display.

• Version 0.91.02: maintenance revision.

• Version 0.91.01: Major version release.

  This corrects a few minor bugs including crashing when starting up an empty database (that bug was introduced sometime in the last month). It is the first release with some reorganized code.

• Version 0.90.08: Fixed bug in "Filter by 'Good Spot data'" where the initial "Check spots for Good Spot mode" was not properly initialized.

• Version 0.90.07: Additional fixes in QualCheck mapping now handles the three types: Alphabetic codes, MAExplorer integer property codes, continuous floating point monotonic quality value. If the latter is used, then a "Spot Quality threshold" slider will popup when it is invoked. The MAExplorer Plugins Web page is also available at thei Web site and will track progress with the development of the MAExplorer Plugins Open Java API for investigators adding their own analysis methods.

• Version 0.90.06: Fixed bugs in Genomic database popups. It was ignoring the request to popup a web page for particular genomic IDs.

• Version 0.90.05: Added QualCheck mappings of Affymetrix "Abs Call" of "P" (or "G" or "T") to Good Spot, "A" to Bad Spot, and "M" (or "B" or "F") to Marginal Spot. This allows the "Filter by 'Good Spot data'" to handle this type of data.

• Version 0.90.04: Added experimental GenomicMenu/URL configuation options for future use.

• Version 0.90.03: Added QualCheck data filter for "Filter by 'Good Spot data'". This will be active if the QualCheck data is available in the .quant files.

• Version 0.90.02: Fix bug in some of the State scroller controls when adjusting a scroller latches into a fixed state that can only be changed by restarting MAExplorer. Added LocusLink popup Web pages from LocusID (if it exists) or LocusLink from GenBank identifiers.

• Version 0.90.01: Major version release.

  (very Beta) Changing convention so use "MasterID" as the master gene index. This makes it more flexible than when used the CloneID as the master gene index. Added LocusLink.

• Version 0.89.41: Added new scatter plots for comparing Cy3/Cy5 ratio data to compare either Cy3 or Cy5 of HP-X against Cy3 or Cy5 for HP-Y. Added additional change limit constraints "Compare channels meeting range" to "Filter by spot intensity [SI1:SI2] sliders". This allows partial filtering by spot intensity.

• Version 0.89.40: Some of the State Scrollers were too sensitive at low values. Therefore, we set them to respond non-linearly with a more precise vernier at the low end.

• Version 0.89.39: Fixed bug t-Test if using multiple samples with arrays with 1 Field.

• Version 0.89.38: Fixed bug in Filter that allows the user to use either "positive only" or "positive or negative" intensity data (used with Affymetrix data).

• Version 0.89.37: Increased the dynamic range of the scrollers for p-value, cluster distance, [Z1:Z2], |Diff XY|. Fixed labels in the Cy3 vs Cy5 plot so they are updated when the current sample changes.

• Version 0.89.36: Fixed error in "Inside Range" for "Filter by Ratio ..." and "Filter by Cy3/Cy5 Ratio ...". Fixed HP-X,Y,E labels in the popup HybProbe menu "Choose HP-X,Y,E ..." command. Also added a new button "<< that removes all entries from the Selected list to the Remainder list. Changed the way the Cy3 vs Cy5 (for ratio data) or F1 vs F2 (if you are using intensity data where all spots are replicated) scatter plots work - the HP-X vs HP-Y scatter plots are not changed. The data Filter still works for all of the plots. Changed the name of the "NPN clustering" method in the menus and reports to "K-means clustering" since that is essentially what it is.

• Version 0.89.35: Improved HP sample ID reporting in various lists, lists, etc. Added HybProbe menu command "Edit use (Cy5/Cy3) else (Cy3/Cy5) for each HP" for use with ratio data. This selectively swaps (Cy3,Cy5) data entries so may use (carefully!) dye-swap data for replicates. Added UP/DOWN buttons in the HP chooser to make it easier to adjust the order of the HP-E list.

• Version 0.89.34: Changed the pseudoarray image generation paradigm so that it uses a fixed size spot and spacing and makes the pseudoarray image size dependent on the number of grids, rows, columns and fields. This results in a more consistent interface.

• Version 0.89.33: Added "Filter by Cy3/Cy5 HP-X ratio or Zdiff sliders" to the Filter menu. This is useful for filtering data from a single sample.

• Version 0.89.31: Added Rename command for Gene Sets and Condition lists. Added "SaveAs GeneSets" to K-means clustering. This saves all of the clusters as named Gene sets ("Cluster #1", "Cluster #2", etc.);

• Version 0.89.30: Fixed problem with popup Web browser for Macintoshes.

• Version 0.89.29: Optimized sorting on startup so it starts much faster when working with large arrays.

• Version 0.89.28: Gene set and Condition set operations now let you select the sets by a pull-down choice menu as well as by typing in names of sets. Additional error checking added for input data files.

• Version 0.89.26: Enhanced configuration file parser and sample name menu default; better error checking on input files to help catch some errors in manually edited user data files.

• Version 0.89.25: Fixed configuration parser errors when reading non-standard data.

• Version 0.89.24a: Fixed data for MAExplorer MGAP demo.

• Version 0.89.24: Fixed bugs in popup browser to use GenBank ID (if it exists) with NCBI server, else it uses the Clone-ID (if it exists) with the nciarray cloneID-to-GenBank server.

• Version 0.89.22: Fixed bugs in ClusterGram, saving current Gene Class.

• Version 0.89.21: Fixed bugs in HP Chooser, Pseudocolor Red-Yellow-Green HP-XY plot menu, HP vs. HP correlation Report.

• Version 0.89.20: Added correct scale for (Plot | Show Microarray | Pseudocolor Red-Yellow-Green HP-XY ratio or Zdiff).

• Version 0.89.19: Added (Plot | Show Microarray | Pseudocolor Red-Yellow-Green HP-XY ratio or Zdiff) command to display the X and Y data as an additive sum of red (HP-X) and green (HP-Y) data.

• Version 0.89.18: New version of download Java installer program that improves portability on different platforms (eg. MacOS, etc). Minor changes (spelling errors in parts of the program). Improved documentation.

• Version 0.89.17: Added commands and functionality to a) find all copies of a named gene in the guesser using the SetE.G.L. button, b) added the "Replicate genes" GeneClass, c) added the "Filter by Genes with replicates" data Filter.

• Version 0.89.16: Fixed several consistency checks when reading non-standard array data.

• Version 0.89.15: Ignore enclosing space characters in SamplesDB, Configuration and .mae startup files.

• Version 0.89.14: Fixed bug in Gene guesser when press the "Done" button. It sets the gene, but did not update the pseudoarray image. Disable genomic browser options in View menu if dbEST3'/5' or GenBankAcc3'/5' is not in the GIPO file database. Since GenBank and dbEST may be accessed via the mAdb Clone report or UniGene report, this should not be a problem. In the Reference Manual a) figures were updated for Genes instead of the older notation of Clone (Version 0.88.*); b) the Short Tutorial (Appendix A) was updated and clarified.

• Version 0.89.12: Added "/Report" directory to hold SaveAs text (.txt) and plot window image (.gif) files. Improved fatal error (called DRYROT errors) reporting. Additional parts of the state saved.

• Version 0.89.11: Added "Filter by positive quant. data" to filter out raw quantified data with negative values which may occur with some types of arrays. Also improved some of the popup window updates when the data Filter, normalization, EGL, or current clone changes.

• Version 0.89.10: Fixed bug in correlation coefficient computation so it works correctly with all Gene Classes and data Filter.

• Version 0.89.09: Fixed bug in reading configuration file that occurs if SubMenus are omitted. Fixed bug recently introduced when updating the finding clones similar to the current clone.

• Version 0.89.08: Fixed bug in auto-update of List of Gene Sets when used with adding(removing) genes using the Control(Shift) clicking on a gene in the array image, scatter plot, etc.

• Version 0.89.07: Fixed bugs in state saving.

• Version 0.89.06: Save the full State when do a (File | SaveAs DB). The full state is restore when you startup the database you saved. Corrected bug in Gene Set Difference operator. Clarified some menu option names. Can now set # genes to report or Filter in a highest/lowest ratio genes in (Filter | Set max # genes in highest/lowest report). Added Report for (Report | Genes in 'Normalization Gene List').

• Version 0.89.05: Added information fields in HP chooser.

• Version 0.89.04: Corrected bug in Ratio Median correction option if using ratio data (i.e. Cy3/Cy5). Improved scatter plot labeling.

• Version 0.89.03: Optimized the main screen area usage. Corrected data-dependent bug which sometimes occurs with the Use Ratio Median correction option if using ratio data (i.e. Cy3/Cy5).

• Version 0.89.02: Set the Use Ratio Median correction option if using ratio data (i.e. Cy3/Cy5).

• Version 0.89.01: Major version release.

  Because MAExplorer can be used with both spotted clone arrays and oligo arrays,we renamed clones as genes (except where Clone ID is used) in both the MAExplorer program and in the Reference Manual.

• Version 0.88.08: Enabled saving the name of the last project when you exit so it knows which file to use as a default when you do an "Open disk DB". Fixed the mnX/mnY data in XY-Statistics Report. Fixed title update in main window when open a new database. Also has fix for Linux Java installer (tested on Red Hat Linux version 6.0).

• Version 0.88.07: Disabled the (File | Set Project) command as there is occasionally a problem with reading the maeProject.txt database file which causes it to hang. This should not be a problem for users with multiple databases as you still get the popup startup file browser when you invoke (File | Databases | Open file DB).

• Version 0.88.06: Edit submenu "Sets of Conditions" HP lists is working and is saved/restore with the File menu "Save disk DB"/"Open disk DB".

• Version 0.88.05: Added mouse-over to main window and moved gene locator button to left. Added "Use ratio median correction" option to Normalization menu. This option can be used for rescaling Cy3, Cy5 intensity data if the medians are unequal. Also, the manual has been updated and more discussion on data mining has been added.

• Version 0.88.04: Edit submenu "Sets of HybridProbes" Condition lists are partially working (not working with File menu "Save disk DB"/"Open disk DB", set operations not enabled yet).

• Version 0.88.03 Edit submenu "Sets of clones" and File menu "Save disk DB"/"Open disk DB" Save/Restore clone sets as .cbs clone bit set files in stand-alone.

• Version 0.88.02: cleaned up SamplesDB.txt and MaExplorerConfig.txt specifications and sample files in Appendix C.

• Version 0.87.08: Fixed some bugs in HP-X,-Y,-E chooser. Added additional checking for missing field names in data files. If a field is incorrect, MAExplorer can't continue. It now gives a popup "Dryrot" error message to give us information about which fields are incorrect.

• Version 0.87.06: May now use combined chooser for selecting the active HP-X, HP-Y and HP-E probes.

• Version 0.87.05: May now save all popup text windows as .txt files in stand-alone mode.

• Version 0.87.01: May now save all plot windows as GIF images as .gif files in stand-alone mode.

• Version 0.86.36: Fixed bug in "Filter by spot intensity [SI1:SI2] sliders". Allow simultaneous viewing of HP-X vs HP-Y and Cy3 vs Cy5 (or F1 vs F2) scatter plots. You may now also simultaneously view and HP-X/HP-Y and Cy3/CY5 ratio histograms.

• Version 0.86.35: A new "Filter by spot intensity [SI1:SI2] sliders" mode has been added to the Filter menu. It is useful for filtering individual changes when analyzing Cy3/Cy5 data.

• Version 0.86.34: The new stand-alone distribution is now generated with InstallAnywhere 3.5 instead of 3.5 beta. This corrects some of the problems we had seen with very large core images.

• Version 0.86.34: The handling of "Out of memory errors" seen when clustering large numbers of clones on a small computer has been changed. Previously, it shut off the cache and continued the computation - although VERY slowly. Now, it prints out the following message:

   There is not enough memory to cluster current filtered clones. Options:
      1. reduce the number of filtered clones and try again, or,
      2. disable cluster-cache (Clustering menu) - will be VERY slow.
  

• Version 0.86.33: A new "Set project" command has been added to the File menu. It selects the current project, if multiple projects have been set up (using the future "New project" command).

• Version 0.86.29: A new "Show mouse-overinfo" mode has been added to the View menu (default is on). It reports HP probe and spot/clone details in the array image, scatter plot, expression profile overlay plot, etc.

• Version 0.86.27: In stand-alone mode, the "Open disk DB" now lets the user open a .mae startup file. The user browses the local file system to select the startup file to use.

• Version 0.86.25: When using ratio (i.e. Cy3/Cy5) data, the menu entries, plot and report legends change the names from (F1,F2) to (Cy3,Cy5) where appropriate.

• Version 0.86.24: Corrected bug in using the ratio and intensity thresholding of mean HP data was previously using the single probe data.

• Normally, MAExplorer works well as an applet with Netscape 4.7 or Windows Internet Explorer 5.0 on Windows PCs and on SUN Unix. It does not seem to work well as an applet on Macintoshs and problems have been reported on SGI systems.

• One solution is install Sun's HotJava browser. The HotJava browser is available from Sun Microsystems . You can download HotJava for the Windows-95/-98/-NT/-2000/-XP and for Solaris (Unix).

• Reguardless of which browser is used, because it is is doing a lot of computation, MAExplorer runs best on a fast machine with lots of memory. This is especially true when running a browser because of the additional browser overhead.

Figure 4.4 Example of a fatal Dryrot Error window. This may occur for a variety of reasons. This window lists the main reason and also lists some of the MAExplorer state information. If you wish, you may save this window (press the "SaveAs" button) and mail it to us. We may try to correct the problem in the next release if it is a problem with MAExplorer. Alternatively, it could be a user data error.

Figure 4.4.1 Example of a fatal Dryrot Error window after SaveAs. This tells you where the saved error message file was saved and the email address to send it to if you wish.

Release Archive for stand-alone MicroArray Explorer on NCI/LECB

Acknowledgements

Primary contributers to Cvt2Mae were Peter Lemkin (LECB/NCI), Greg Thornwall (SAIC/FCRDC), Bob Stephens (ABCC/NIH).

We wish to thank the many members of Lothar Hennighausen's Laboratory of Genetics and Physiology (NIDDK) who inspired the initial development of MAExplorer and its continued development. Thanks also to:

Greg Alvord (SAIC/FCRDC),
Kevin Becker and Chris Cheadle (NIA/NIH),
Breast Cancer Think Tank (NCI),
Damien Chaussabel (NIAID),
Terry Clark and Josef Jurrek (U. Chicago),
Mitko Dimitrov (LECB/NCI),
Jai Evans and Chris Santos (DECA/CIT/NIH),
Troy Moore (Research Genetics),
Peter Munson (CIT/NIH),
Alan Li (SourceForge),
Quang Tri Nguyen (LECB+LCRC/NCI),
John Powell and Esther Asaki (CIT/NIH),
Eric Shen (U. Arizona),
Moshe Shani (Agr. U. Israel),
Richard Simon (NCI/NIH),
Bob Stephens and Gary Smithers (ABCC/FCRDC),
Ron Taylor (U. Colorado),
Mark Vawter (NIDA/NIH, UC-Irvine),
John Weinstein (LMP/NCI), David Kane (SRA/NCI), Ajay (LMP/NCI),
and to many others for useful discussions and suggestions that have helped improve the MAExplorer's capabilities and usability.

Thanks also to Jeff Thomas, Charmaine Richman, and Tom Stackhouse (NCI) for helping with the MAExplorer Open Source process.

References to related exploratory data analysis methods and MAExplorer

1. Beardsly T (1999) Scientific American, March 1999, 35-36.

2. Bickel, D (2001) "Robust Cluster Analysis of DNA Microarray Data: An Application of Nonparametric Correlation Dissimilarity. Proceedings of the Joint Statistical Meetings of the American Statistical Association (Biometrics Section, to appear). [ http://www.mathpreprints.com/math/Preprint/bickel/20010730/3/]

3. Brazma A, Hingamp P, Quackenbush J, Sherlock G, Spellman P, Stoeckert C, Aach J, Ansorge W, Ball CA, Causton HC, Gaasterland T, Glenisson P, Holstege FC, Kim IF, Markowitz V, Matese JC, Parkinson H, Robinson A, Sarkans U, Schulze-Kremer S, Stewart J, Taylor R, Vilo J, Vingron M (2001) Minimum information about a microarray experiment (MIAME) - toward standards for microarray data. Nature Genetics 28: 365-371. [See http://genomics.nature.com/supplementary_info, and http://www.mged.org/)]

4. Cleveland WS (1985) The Elements of Graphing Data. Wadsworth Press, Monterey, CA, pp 1-323.

5. DeRisi J, Penland L, Brown PO, Bittner ML, Meltzer PS, Ray M, Chen Y, Su YA, Trent JM (1996) Use of a cDNA microarray to analyze gene expression patterns in human cancer. Nature Genetics 1996, 14: 457-460.

6. Dudoit S, Yang YH, Callow MJ, Speed TP (2000,Aug) Statistical methods for identifying differentially expressed genes in replicated microarrays experiments. TR-578, Dept. of Statistics, Stanford Univ., Stanford, CA.

7. Eisen MB, Spellman PT, Brown PO, Botstein D (1998) Cluster analysis and display of genome-wide expression patterns. PNAS USA 95: 14863-14868.

8. Jagota, A (2001) Microarray Data Analysis and Visualization. Bioinformatics By The Bay Press, 101 pp. ISBN-09700297-3-X.

9. Kaufman L, Rousseeuw PJ (1990) Finding Groups in Data: An Introduction to Cluster Analysis. J. Wiley and Sons, Inc., New York.

10. Kerr MK, Churchill GA (2001) Statistical design and the analysis of gene expression microarray data. Genet Res. 77(2):123-8. Review.

11. Kerr MK, Churchill GA (2001) Bootstraping cluster analysis: Assessing the reliability of conclusions from microarray experiments. PNAS 98:8961-8965.

12. Korn EL, Troendle JF, McShane LM, Simon R (2001) Controlling the number of false discoveries: Application to high-dimensional genomic data. BRB, NCI, Bethesda, MD. Aug 22, 2001, Tech Report #3.

13. Lipkin LE, Lemkin PF (1980) Data-base techniques for multiple two-dimensional polyacrylamide gel electrophoresis analyses. Clin. Chem. 26: 1403-1412. (GELLAB-II home page http://www.lecb.ncifcrf.gov/lemkin/gellab.html)

14. Lemkin PF, Lester EP (1989) Database and search techniques for two-dimensional gel protein data: a comparison of paradigms for exploratory data analysis and prospects for biological modeling. Electrophoresis 10: 122-140.

15. Lemkin PF (1995) Representations of protein patterns from 2D gel electrophoresis databases. In: Pickover, C., (Ed) The Visual Display of Biological Information, World Scientific Publishers, River Edge, New Jersey, pp 43-59.

16. Lemkin PF (1993) Xconf: a network-based image conferencing system. Comput. Biomed. Res. 26, 1-27.

17. Lemkin PF (1997) Comparing Two-Dimensional electrophoretic gels across the Internet. Electrophoresis, 18: 461-470. [Extended paper], (Web site http://www.lecb.ncifcrf.gov/flicker/)

18. Lemkin PF, Thornwall G (1999a) Flicker image comparison of 2-D gel images for putative protein identification using the 2DWG meta-database. Molecular Biotechnology, 12:(2) 159-172. [Extended paper and Web site http://www.lecb.ncifcrf.gov/2dwgDB/]

19. Lemkin PF, Myrick JM, Lakshmanan Y, Shue MJ, Patrick JL, Hornbeck PV, Thornwall GC, Partin AW (1999b) Exploratory Data Analysis Groupware for Qualitative and Quantitative Electrophoretic Gel Analysis Over the Internet - WebGel. Electrophoresis 20:(18) 3492-507. (PDF), (also see WebGel server: http://www.lecb.ncifcrf.gov/webgel).

20. Lemkin PF, Thornwall GC, Walton KD, Hennighausen L (2000) The Microarray Explorer tool for data mining of cDNA microarrays - application for the mammary gland Nucleic Acid Research 28:(22) 4452-4459. (PDF).

21. Lemkin PF, Thornwall GC, Walton KD, Hennighausen L (2000) MicroArray Explorer - a tool for data mining of microarrays - Overview. (PDF).

22. Lemkin PF, Thornwall GC, Walton KD, Hennighausen L (2000) MicroArray Explorer - a tool for data mining of microarrays - Examples. (PDF).

23. Lemkin PF, Thornwall GC, Powell J, Asaki E (2000) Using MAExplorer with the NCI/CIT mAdb Web server. (PDF) and [http://nciarray.nci.nih.gov/]

24. Lemkin PF (2001) Introduction to Data Mining of Microarrays using the MicroArray Explorer. (PDF) or (PPT).

25. Lemkin PF (2001) Using Cvt2Mae to convert Affymetrix array data for MAExplorer. (PDF)

26. Lemkin PF, Thornwall G, Hennighausen L (2001) MicroArray Explorer - A Java-based Tool For Data Mining Microarrays. Paper presented at the AMS-IMS-SIAM Summer Conference on Statistics in Functional Genomics, June 10-14, 2001. (PDF).

27. McShane LM, Radmacher MD, Freidlin B, Yu R, Li M-C, Simon R (2001) Methods for assessing reproducibility of clustering patterns observed in analyses of microarray data. BRB, NCI, Bethesda, MD. July 1, 2001, Tech Report #2.

28. Radmacher MD, McShane LM, Simon R (2001) A paradigm for class prediction using gene expression profiles. BRB, NCI, Bethesda, Nov 19, 2001, Tech Report #1.

29. Schneiderman B (1997) Designing the Human Interface, 3rd edition. Addison-Wesley Pub. Co., NY, pp 1-638.

30. Schulze A, Downward J (2001) Navigating gene expression using microarrays - a technology review. Nature Cell Biology 3: E190-E195.

31. Simon R, Radmacher MD, Dobbin K (2001) Design of Studies Using DNA Microarrays. BRB, NCI, Bethesda, MD. 2001, Tech Report #4.

32. Sneath PHA, Sokol RR (1973) Numerical taxonomy. W.H. Freeman and Co, San Francisco. pp 1-573.

33. StatSoft Inc. (2002) Electronic Statistics Textbook, Tulsa, OK: StatSoft. Available at http://www.statsoftinc.com/textbook/stathome.html

34. Tufte E (1997) Visual Explanations. Images and quantities, evidence and narrative. Graphics Press, Cheshire, CT, pp 1-156.

35. Tukey J (1977) Exploratory data analysis. Addison-Wesley Pub. Co., Reading, MA, pp 1-688.

36. Vawter MP, Barrett T, Cheadle C, Sokolov BP, Wood WH III, Donovan DM, Webster M, Freed WJ, Becker KG (2000) Application of cDNA Microarrays to Examine Gene Expression Differences in Schizophrenia. Submitted.

37. Weinstein JN, Myers TG, O'Conner PM, Friend SH, Fornace AJ, Kohn KW, Fojo T, Bates SE, Rubinstein LV, Anderson NL, Buolamwini JK, van Osdol WW, Monks AP, Scudiero DA, Sausville EA, Zaharevitz DW, Bunow B, Viswanadhan VN, Johynson GS, Wittes RE, Paull KD(1997) An Inforation-Intensive Approach to the Molecular Pharmacology of Cancer. Science 275: 343-349.

38. White KP, Rifkin SA, Hurban P, Hogness DS (1999) Microarray Analysis of Drosophila Development during Metamorphosis. Science 286: 2179-2184.

There is also a pre-computed example of an Ordered Condition List using 4 conditions of replicates of C57B6 (pregnancy day 13, lactation days 1 and 10, and stat5a(-,-) 15 samples. The database also includes 4 additional condition sets of this data and an Ordered Condition List of the 4 conditions (in the State/ directory). This may be used to demo the OCL F-test filter.

If you have access to another MAExplorer database, you can use it instead since the tutorials are fairly generic.

Using the stand-alone application for the tutorial

These same subsets as well as other subsets of the MGAP data are available in the set of .mae startup files distributed with MAExplorer. To access these files,

1. Start MAExplorer after you have installed it. Eg. in Windows, go to the Windows "Start Menu" and click on MAExplorer. If it is not in your Start Menu, you can go to where you installed it (typically C:\Program Files\MAExplorer) and click on MAExplorer.exe.
2. Then after it starts, go to the "Files" menu and select "Open disk DB" and select the startup file you want. Alternatively, you can go directly to the list of startup files in C:\Program Files\MAExplorer\MAE) and double-click on one of the startup files.

A.2 General instructions:

• As a stand-alone application, select the startup file entry (files ending with a ".mae" file extension) from a directory of startup files on your local computer. Generally these are in a subdirectory called MAE in a project directory (see Appendix C. Use of MAExplorer with other microarrays).

When it starts, a main window will pop up. It then downloads a gene database tables and the particular hybridized samples you specified. When it is ready for you to begin interaction, the menu bar will become active and it will display a green Ready - click on a gene to query database message. Depending on your Internet connection speed, it may take a few minutes to set up. If you are running MAExplorer as a stand-alone application and it is getting data from your local disk, startup will be much faster.

Second, go to the A.3 instructions for self-guided tutorial below for instructions on what to do next.

HINT: print this tutorial page and then read the following instructions from the printout rather than trying to keep this window visible. You might also print the parts of the MAExplorer Reference Manual for the same reason.

HINT: You might want to keep a record of the commands you have used or the messages and measurements you have made. To do this you need to enable message and command history logging. Go to the View pull-down menu and then select the type of logging you want using the Show log of messages or the Show log of command history commands.

NOTES:. On computers with low resolution (i.e. less than 1024 X 780) you may need to resize the windows and move them to different parts of the screen to view them simultaneously.

1. ratio between two conditions X and Y (HP-X, HP-Y)
2. expression profile of a set of conditions (HP-E) (see Example A.3.1.7)

Note of caution: if the signal is close to background the X/Y ratio may be bogus.
You can filter out low intensity genes by

1. Analyze expression of individual genes
2. Analyze expression of gene families and clusters
3. Compare expression patterns in multiple hybridized samples

Here are some things to try

1. You could click on the Reference manual in the Help menu. That is this reference manual, but it appears in a new Netscape Web browser window so you may read it while working on MAExplorer. Delete the window when you are finished with it.

2. Click on the Glossary in the MAExplorer Help menu. This section describes terms used in MAExplorer. Delete the window when you are finished with it.

3. Click on the Index. It may be useful in finding things that are not in the table of contents.
4. Note: a hybridized sample is a microarray hybridized with cDNA derived from the mRNA from a particular experiment sample (see notation defined in the Overview). Currently, you may start MAExplorer with preloaded set of samples by starting the stand-alone MAExplorer on a .mae file. Alternatively, you may start it with no samples loaded. It the latter case, you would load samples you are interested in from the Samples menu.

1. Click on different genes (i.e. spots in the pseudoarray image). The pseudoarray image may or may not correspond to the actual array - depending on how the data was derived. Notice the data that gets displayed in the three text lines above the image. If the spot you click on is a named gene (e.g. [1-A4,3] at row 4 column 3 in Field 1), it will also print out the GeneName of the gene.

2. Look at the pull-down menus. They consist of sets of commands with similar functionality grouped together in sub-menus. In particular, look at the Analysis menu. It contains an ordered list of submenus that may be thought of as the sequence one might perform an analysis. In reality, an analysis is more complicated and involves iterating various steps (see Figure 3.1).

3. Go to the "Samples" menu. This menu allows selecting hybridized samples for the HP-X, HP-Y, HP-X and HP-Y 'sets', and the HP-E list. You have several ways to do this. The easiest way is to use the "Choose HP-X, HP-Y and HP-E" sample selection wizard. The other alternative is to use one of the set of cascading menus that may be used to change the selected hybridized samples. Note that the HP-X and HP-Y submenus assign samples for subsequent analysis such as X-Y scatter plots. The HP-E submenu sets the list of samples for expression profiles. Note how you may assign a sample either by going through the cascading menus or from an alphabetic list of all hybridized samples in the pseudoarray image. To change the default HP-X (HP-Y) sample, click on the purple "[X]" ("[Y]") box on the left side of the image (above the list of samples) so it is selected. Then click on the desired purple "*" adjacent to the samples listed on the left edge of the pseudoarray image. You may switch between using single and multiple samples (i.e. 'sets') with HP-X and HP-Y. Note the "HP-X:" and "HP-Y:" labels at top left when you switch between single and multiple samples. Go to the HP-X/-Y 'set' and HP-E submenus and list the contents of the respective sets to see what they contain. Note how one may add or remove samples from these sets.

4. Go to the "Samples" menu. Then select "Choose named condition list of samples". This lets you define new condition lists of samples. Go to the "Edit" menu then "Sets of Conditions (samples)" to see additional ways to manipulate these condition sets. For example, you might define a new condition and the assign it to the working HP-X 'set'.

5. Go to the "Samples" menu. Then select "Choose ordered list of conditions". This lets you define a new or edit an old Ordered Condition List(OCL). In the included MGAP there, there is a pre-computed example of an Ordered Condition List using 4 conditions of replicates of C57B6 (pregnancy day 13, lactation days 1 and 10, and stat5a(-,-) 15 samples. The database also includes 4 additional condition sets of this data and an Ordered Condition List of the 4 conditions (in the State/ directory). This may be used to demo the OCL F-test filter. Go to the " Filter" menu and select "Filter by current Ordered Condition List (OCL) F-test [p-Value]". This will popup a p-value slider where you can adjust the criteria for selecting genes passing the F-test.

6. Go to the "GeneClass" submenu in the "Analysis" menu. This is set of cascading menus that may be used to change the default Gene Class. Different genes belong to different gene classes and this is a way of sub-setting the data. You may currently set it to All Genes, All Named Genes, ESTs similar to genes, ESTs, Good genes, All named genes and ESTs, Replicate genes (multiple copies on the array), Calibration DNA. Select "ESTs similar to genes". Notice that the display now only shows the red (white) circles on the ESTs similar to genes in the intensity (ratio) pseudoarray image. Look at the circle overlays on the microarray image - note how they changed. Go back and set the gene class to "All genes" and then "Calibration DNA". The "Calibration DNA" is the set of spots that may be used for normalizing data between microarrays. Check out the "Set gene class subset" submenu. Leave the gene class set to All Genes.

7. If you are connected to the Internet, go to the "View" Menu. Then turn on the switch "Enable display current gene in popup XXXX Web Browser". Depending on your database, XXXX may be GenBank, LocusID, UniGene, dbEST or mAdb Clone DB. Then click on a gene in the image. It pops up a Web browser showing the genomic database Web page for that gene (if any). If you click on a different spot, it will reuse the popup Web browser with new data. If you don't want it to be active, go back to the "View Menu" and click on "Enable display current gene ..." again to disable it.

8. Go to the "Normalization" submenu in the "Analysis" menu. This is used for normalizing data between microarrays so they may be compared. The default is to scale the data to the median value for each array. Investigate the other normalization methods. If your quantified data contains spot background data, you may also enable background correction that subtracts a microarray specific overall background intensity value from each intensity value in each array.

9. Go to the "Report" submenu in the "Analysis" menu. You may generate a report in several formats "Spreadsheet" or "tab-delimited", the latter being cut and paste compatible with Excel. In the "Samples Report" sub-submenu in the "Report" submenu, click on "Hybridized Samples", then switch to the other Report format and do it again. Click on "Hybridized Sample Web links". In the "Gene Reports" submenu, try "All named genes". Then try the "Highest HP-X/HP-Y ratio" in the "Filtered gene reports" submenu. If you are using a list of HP-E or HP-X/-Y 'sets' of samples, you might try looking at the expression profile ratios or statistical data respectively though other Reports.

10. Review the data " Filter" submenu in the "Analysis" menu. Select the "Filter by expression sliders" [I1:I2]. Expression is intensity in the case of ³³P or biotin labeled, or (Cy3/Cy5) in the case of ratio data. This pops up a state sliders window. The I1 scroll bar (lower limit of the gray value intensity) to about 100. Look at the genes that were eliminated because they are out of the range. If you select the "Filter by spot intensity sliders" [SI1:SI2], it will filter genes by spot intensity in either F1 or F2 duplicate (if you have duplicate spots), or the Cy3 or Cy5 spots. You can select which sets of HPs to use (current HP, HP-X and HP-Y, HP-XY sets, or HP-E). You might disable the [I1:I2] filter while you experiment with [Si1:SI2] filter. If you have (F1,F2) or (Cy3,Cy5) data, put up the F1 vs F2 or Cy3 vs Cy5 scatter plot and you can visualize how the thresholding works.

    In the Filter menu, add the "Filter by ratio or Zdiff sliders". Then the [R1:R2] ratio range sliders are added to the state slider window and may be used for filtering genes. If the normalization method is one of the Zscore methods, it filters by the difference of the Zscores otherwise by the ratio and the [Z1:Z2] range is used. Note that the genes that pass the filter will appear to have a red (white) circle in the pseudoarray intensity (ratio) grayscale (pseudocolor), or red "+" in the scatter plots so you might try moving the controls while in those plot modes. Try some of the other filters. The spot CV test removes genes where replicate spot values (F1 and F2 in the case of a single sample or replicate samples in the case of HP-X and HY-Y 'sets' or the HP-E' list of genes) are not well correlated. The t-Test filter may be used with sets of X and Y samples to find genes with a p-value less than the specified threshold.

11. Go to the "Plot" submenu in the "Analysis" menu. Then got to the "Show Microarray" submenu, try pseudocolor ratio (or Zscore) modes and finally leave it in the pseudograyscale image intensity mode. Try using the "Scatter Plots" and "Histograms" in the Plot menu. Vary the normalization methods and see how it affects the array image and scatter plots. When you are in a scatter plot, click on a point. It will display data similar to when clicking on an image. If UniGene data is available in your database, Go to the Views menu and set "Enable display current gene in popup UniGene Web browser" and click on a point again. This pops up another Web browser window and lookup that gene in the UniGene database. Change the threshold slider and notice how points appear and disappear. Click on a bin in the ratio histogram, it will filter the genes so that only the genes that have ratios in that bin are displayed.

12. Go to the "Plot" submenu in the "Analysis" Menu and then to the "Expression profile" submenu. Then select the "Display gene expr. profile for HP-E". It pops up a window with two buttons "Show HP names" and "Close". Then click on a gene in the image. It will draw the expression profile for that gene. Move the mouse so it is over one of the vertical bars in the plot to get the data for that particular HP. If you click on a different spot in the image it will display the new expression profile in the same window. Click on the "Show HP names" button to popup a window with the list of HPs matching the numbers in the expression profile plot. Now create a scatter plot of two HPs and then click on a red "+" in the plot. It will update the expression profile plot with this gene. If you want to compare several expression profile plots, repeated create the "Display gene expression..." windows from the View menu. Move the windows close to each other so they are easier to compare. Only the last one you created allows you to change the gene. If you don't want a popup window anymore, click on its "Close" button. You may save the scatter plot as a GIF file by pressing the "SaveAs" button which will save it in the Report subdirectory associated with the startup data..

13. Next we look at gene clustering. Go to the "Plot" submenu in the "Analysis" menu, and then to the "Cluster plots" submenu, try "Cluster genes with expression profiles similar to current gene". This pops up a slider with the cluster threshold. Then click on a gene in the image. It will then popup a text window with the genes whose cluster distance is less than the cluster threshold. It is sorted by minimum cluster distance. Notice that some genes in the image have different size blue boxes around them. The larger the box, the smaller the cluster distance and the more similar. Move the cluster threshold slider. This will change the clustering as seen in both the image and in the cluster popup window. Click on the "Report" button in the text window. This will popup a report on these genes sorted by minimum cluster distance. Then click on the "EP plot". This pops up a scrollable list of expression profile plots for the genes you have filtered by this test on so you can review the actual expression profiles. Close this window and set the Filter to a small number of genes such as with "ESTs similar to genes". Note that the genes passing this filter are saved in the E.C.L. and may be saved as a gene subset or part of the data Filter.

    Turn on one or more Filters to reduce the number of genes to say under 100 (e.g. t-test or spot CV filters). Then press the "Go 'Cluster all genes'" button in the cluster window. This is equivalent to invoking the "Cluster counts of Filtered genes by expression profiles" command from the "Cluster plots" submenu. Notice the Filtered genes has blue circles of different sizes. The larger the circle, the more genes there are that are similar to that gene. Move the cluster threshold slider and note that the number of similar genes changes, the size of the blue circles will change. As with the other cluster mode, you may generate a report of sorted cluster counts. Click on a gene with the largest green circle. This will then switch you back to single gene clustering mode where you can investigate that gene in more detail.

14. Next we look at K-means gene clustering. Go to the "Plot" menu and in the "Cluster plots" submenu, try 'Display K-means gene expression profiles for Filtered HP-E' (similar to K-means clustering). This pops up a scroller for "N-clusters", the number of clusters desired, with a default of 6 clusters. It will then popup a K-means report window with various controls. The centers of the N clusters are indicated with magenta circles where the size corresponds to the number of genes in the cluster. If you click on a gene in the array, it will draw all members of its cluster as green numbers (try this with the scatter plot present as well). Press "EP plot" to scroll through a list of expression profiles of the genes sorted by cluster. Press "Mean EP plot" to scroll through the summary expression profiles of the clusters. Press "Cluster-Report" and "Mn-Cluster-Report" to generate reports for these clusters. You may change the seed gene by changing the current gene (click on a different gene in either the array image or the EP plot. Then and press "Recompute" to recompute the clusters. Pressing the "Show HP names" pops up a list of the samples used in the expression profile. Now add a HP-X vs HP-Y scatter plot. If you select a particular gene, it puts all genes associated with the cluster for that gene in the "Current Cluster" and colors them with a green cluster number instead of a "+".

15. Next we look at hierarchical gene clustering. Go to the "Plot" menu and in the "Cluster plots" submenu, try "Hierarchical clustering of expression profiles". Then select "Display clustergram of gene expr profiles". This will compute the clustergram for the Filtered genes with the data normalized by the assigned HP-X sample. It will then popup a clustergram window with various controls. The clustergram is scrollable. You may optionally add the dendrogram with the "Dendrogram" checkbox. Clicking on a row will show data for that gene. Clicking on a box in the clustergram will show data for the particular sample for that gene. You may zoom the dendrogram by repeatedly clicking on the "xxxX DB" zoom button. Press "EP plot" to scroll through a list of expression profiles of the genes sorted the same as the clustergram. Press "ClusterGram Report" to generate a report of the expression profile data sorted the same as the clustergram. Pressing the "Show HP names" pops up a list of the samples used in the expression profile. You may save the entire clustergram - dendrogram image in a GIF file as before by pressing the "SaveAs" button.

16. You may perform operations on sets of genes. For example, merge sets of genes found under two different experiments or conditions. Go to the "Sets of genes" submenu in the Edit menu and pick the "List saved gene sets" selection. This lists the default gene sets. Then select "Assign 'User Filter Gene Set'". This will request a gene set to use with the Filter in a pop up dialog box. Select the set for "Genes in current cluster class" that you saved in the previous example. Then press "Ok" in the dialog box. Then select "All genes" in the GeneClass menu. This resets the filter to look at all genes. Select "Filter by 'User Gene Set' membership" in the Filter menu. This restricts the genes to the saved current cluster in the previous example.

17. Gene set operations may be performed on pairs of gene sets. Select "Union of 2 gene sets" entry from the "Sets of genes" submenu in the Edit menu. This will request 3 gene set names in a pop up dialog box. Select 'Similar ESTs' for the 1st gene set name, select "Genes in current cluster class" for the 2nd gene set name, Enter "Union of similar ESTs and genes in current cluster" for new gene set name. Then press "Ok" in the dialog box. This computes the union of the two gene sets into a new gene set. Then select "Filter by 'User Gene Set' membership" as before. This will reset the 'Use Gene Set' for the Filter in a pop up dialog box. Select the set for "Union of similar ESTs and genes in current cluster" just saved. Try saving other Filtered genes sets and doing other gene set operations.

18. Go to the "Help" menu. The MAExplorer documentation and glossary as well as other MGAP documents are available and will appear in a new popup Web browser.

19. If you are running MAExplorer as a stand-alone application, you may save the state of your analysis (but not the gene subsets at this time). Go to the "File" menu and then "Databases" submenu. Select "Save as file DB". Enter a file name to save your startup state. Then you can restart MAExplorer on this data set at a latter time by clicking on this file (in a windows based system). Saving the database will also save the state of the data Filter, the gene sets and the condition lists. You can see this by listing them in the Edit menu sub-options after you have restarted a previously saved database.

MAExplorer requires a specification of array geometry and quantification information. These are defined in a configuration startup file. The startup file contains the initial list of hybridized samples to be loaded, and other parameters such as the name of the configuration file (if it is different from the default name). A stand-alone application causes the .mae startup file (or the PARAM list in the case of an applet) to be read when it is started. The configuration file contains various defaults. If any of these are specified in the configuration file, the override the built in default values. Values from the .mae startup or applet PARAMs will override the configuration file values. These configuration parameters may be overwritten by arguments in the stand-alone .mae startup files or PARAMs in the Applet startup specifications.

A few additional files are required and are defined in the configuration file. These include: a Gene-In-Plate-Order or GIPO file; a samples database file listing names of the samples available for loading; and a   gene class names file. An optional (but deprecated) extra array information file may be specified to access additional data about samples. Quantified hybridized sample array spot data (Quant files) from each array is put into a separate data file. Note that all data files are tab-delimited files such as may be generated with Excel, relational databases or directly from array spot quantification software.

Hybridized sample arrays must be scanned and then spots quantified using other software. MAExplorer does not do spot quantification from scanned image files. However, MAExplorer can use spot data from a variety of array image quantification programs that generate tab-delimited data files. The data needs to be converted to the MAExplorer schema described in this Appendix.

The derivation of quantified spot data files from hybridized sample arrays is discussed later in this section as are in the quant file data format.

The configuration file is created once for each new array GIPO geometry and database of hybridized samples. It is independent of the number of samples. Configuration parameters include array geometry (# of grids, # of duplicate spots/gene, etc), whether the data is intensity or ratio data (e.g. Cy3/Cy5), etc. The configuration file may also include labeling, quantification dynamic range, default analysis thresholds, mapping of used data file table-field names to expected MAExplorer names for the GIPO and quantification files, additional database-specific pull-down menu plugins, names of gene sets and sample condition lists, etc.

The GIPO file is independent of the number of array samples and describes the mapping between spot position in an array and its gene identification as well as corresponding data such as original plate number, row and column; UniGene ID, GenBank ID, dbEST ID, etc. These files will be described in more detail including how one can create the necessary database files that MAExplorer requires for use with various types of microarray data.

Directory (i.e. folder) structure of stand-alone databases

 (specific database directories and files they contain)
              / Cache
                       / (copies of any data files saved from Web DB access)

              / Config 
                       / MaExplorerConfig.txt
                       / SamplesDB.txt
                       / GIPO-db.txt

              / MAE
                       / (set of startup database files).mae 

              / Images
                       / (set of original or sampled array .jpg images) (optional)

              / Plugins
                       / (optional set of .jar or .class MAEPlugin files)

              / Quant
                       / (set of spot quantified data files).quant

              / Report
                       / (set of .txt and .gif report files generated using SaveAs

              / State
                       / (set of gene set files).cbs and
                       / (set of condition list files).hbl generated using Save DB

Figure C.1 Directory structure of stand-alone databases required by MAExplorer. The "/Config", "/Quant", and "/MAE" directories are required. The /MAE directory is only used with the stand-alone version with .mae files, not for the applet. [When used with an applet, the main path is the path of the download JAR file and .mae files are not used.] The "/Report", and "/State" directories are created by MAExplorer as needed and the user need not create them prior to running MAExplorer. The text reports and plot GIF images are saved in the /Report folder when you "Save" a report or plot. When you "Save" the current database session (File | Databases | Save ...), the gene sets and sample lists are saved in the /State folder for use when you restart MAExplorer on the .mae startup file. The optional "/Cache" directory is only used (and then, only optionally) when downloading data from a Web server. The optional "/Image" directory is only used in there are JPEG images of the arrays provided and their resolution and alignment must correspond to the (X,Y) spot data in the Quant files. The "/Plugins" directory is where the MAEPlugins packaged with MAExplorer are normally kept and where MAExplorer looks when you attempt to load a plugin. Since you can browse your file system, they do not have to appear here.

Examples of some of the database files required by MAExplorer

• http://www.lecb.ncifcrf.gov/mae/MGAP-Array-database/Config/
• http://www.lecb.ncifcrf.gov/mae/MGAP-Array-database/Quant/
• http://www.lecb.ncifcrf.gov/mae/MGAP-Array-database/MAE/

• Example-P33_F1F2
• Example-NAME_GRC
• Example-Cy3Cy5

Additional directories used at run-time

Tools for automating the construction a local stand-alone database

If the field headings in the various user's tables are not the same as that required by MAExplorer, you can easily fix this by adding (Table,Field) mapping entries to your version of the MaExplorerConfig-db.txt file (see mapTF for examples).

Note that the optional Menu_Source_Name entry in the Samples-db.txt file specifies the sub-menu, if any, that the sample will appear in the Samples menu By Source sub-menu.

If the optional extra sample information file is used, then make sure the sample names and database file names are the same, and that there are corresponding rows in each table.

C.1 Creating quantified spot data files from hybridized sample arrays

C.1.1 Color and prefix notation for the following tables: (req), (opt), (future)

C.2 Table of samples that can be loaded into MAExplorer

A typical sample database table might look like:

Sample_ID Project Database_File control 1 breastCancer control1 control 2 breastCancer control2 control 3 breastCancer control3 tumor 1 breastCancer tumor1 tumor 2 breastCancer tumor2 tumor 3 breastCancer tumor3

You may optionally include a Database_ID field. For example:

Sample_ID Project Database_File Database_ID control 1 breastCancer control1 270314 control 2 breastCancer control2 270315 control 3 breastCancer control3 270316 tumor 1 breastCancer tumor1 270317 tumor 2 breastCancer tumor2 270318 tumor 3 breastCancer tumor3 270319

The Database_ID may be useful if there are file length problems on some systems (i.e. MacOS 8-9), we offer the option of using the Database_ID as the file name for the .quant (Quant/ directory) and .jpg (Images/ directory) rather than the Database_File name. For example one could specify "Quant/270314.quant" and "/Images/270314.quant" rather than the default "Quant/control1.quant" and "/Images/control1.quant" names.

The Samples database table includes some required as well as optional fields (see Table C.2.1.1):

Table C.2.1 List of Samples data file table fields. The Samples table lists hybridized samples that are accessible to the user and may be loaded into a database session if they wish. (See Section C.1.1 for option notation.)

Field	Description
(req) Sample_ID	descriptive name of the sample, free text. [Note: an older depricated name is "Membrane_ID"]
(req) Project	that the sample belongs. Used for login protection and grouping of samples
(req) Database_File	name of the .quant spot database file, no spaces. This is the file name for the sample.
(opt) DatabaseFileID	database file ID corresponding to Database_File and Sample_ID. For use with RDBMS Web databases (e.g. experiment id #). NOTE: if you are encoding auxillary data files using this identifier, e.g. sampled array images in the Images/ directory, then this field is required if you want to access those images.

Table C.2.1.1 List of optional Samples data file table fields. These fields may be used for some additional operations. If they are not in the Samples DB table, then the operations will not be available. (See Section C.1.1 for option notation.)

(opt) Menu_Source_Name	Sample SubMenu j that this sample belongs. You could use the word "Default" or leave out this entry if you do not want to use sub menus.
(opt)Orig_File_Name	if applicable. The original file name and sample name if the data was split out from a multiple hybridized sample file.
(opt)Strain	if applicable
(opt) Source	if applicable
(opt) Probe	if applicable
(opt) Stage	if applicable (eg, developmental stage, dose, time point, etc)
(opt) Login	(optional) TRUE if login required with a Web server else blank. This is used primarily with the Applet when interacting with a Web server
(opt) GeneCard_URL	GeneCard ID if applicable
(opt) Histology_URL	(e.g. MGAP) histology DB Web page if applicable
(opt) Model_URL	(e.g. MGAP) mouse model database Web page if applicable
(opt) BGLow	global low value of array background intensity
(opt) BGAvg	global average value of array background intensity
(opt) BGRms	global root-mean-square value of array background intensity

Table C.2.1.2 List of optional Samples data file table fields. These fields are not currently used in any computations but are returned in the Sample Array report in Section 2.4.6.1.

(opt) Contributor	name of researcher submitting the sample
(opt) Contrib_Institute	researcher's organization
(opt) Submission_Date	when submitted
(opt) Exposure	minutes or hours of radiolabel or fluorescent exposure
(opt) Sample_Nbr	internal sample number
(opt) FilterType	name of the array layout
(opt) FilterType_Description	additional description of array layout
(opt) Comments	details describing sample
(opt) Researcher	researcher performing the hybridization
(opt) SampleGrid	serial number of the array or grid or internal laboratory numbering. (Useful if reusing arrays etc)

C.3 Quantified spot data file formats

Alphanumeric mappings for Grid, Grid Row and Grid Column data

The Molecular Dynamics Name_GRC numbering mapping for Grid, Grid Row and Grid Column data

Table C.3.1 List of Quant data file table fields. This specifies the spot quantification data. There may be one or more spots, corresponding to the same gene, on each row. (See Section C.1.1 for option notation.)

Field	Description
(opt) field	field for duplicate genes if using single 'RawIntensity' value/Row
(req) grid	grid name (either A,B,C,... or 1,2,3,... )
(req) grid col	column with in a grid
(req) grid row	row within a grid
(opt+alt) NAME_GRC	(alternative specification of "grid, grid col, grid row").
(req) RawIntensity1	intensity value for field 1. Use this form if there is more than 1 intensity value/row.
(req) RawIntensity2	intensity value for field 2 (required if it exists and for Cy3, Cy5 data)
(req+alt) RawIntensity	intensity value for field 1, if only one field used
(opt) Background1	background intensity value for field 1
(opt) Background2	background intensity value for field 2 (if it exists for F1,F2 data or Cy3, Cy5 data)
(opt+alt) Background	background intensity value for field 1, if only one field used
(opt) QualCheck	quality check for data indicating "bad" spots or genes. Current codes are listed in the Table C.4.2 of QualCheck semantics
(opt) DetValue	spot data detection value quality. This could be the Affymetrix MAS5.0 "Detection p-value" or some other metric correlated with spot detection quality in the range of [0.0 : 1.0]. metrix

Note: If NAME_GRC is specified (eg. for use with ImageQuant-NT data), then the explicit (grid, grow row, grid col) fields are not required. Note: For [G grids, R rows and C columns], this would cover a set of spots in the range [1,1,1] through [G,R,C].

Note: If Cy3/Cy5 double fluorescent labeling is used, then the RawIntensity1 and RawIntensity2 fields may be replaced with Cy3RI and Cy5RI names and the (RawIntensity1, RawIntensity2) fields mapped to (Cy3RI, Cy5RI) in the configuration file mapTF entries (table C.5.4 below). (See Section C.1.1 for option notation.)

Field	Description
(req) Cy3RI	RawIntensity1 value for Cy3
(req) Cy5RI	RawIntensity2 value for Cy5
(opt) Cy3Bkgrd	Background1 value for Cy3
(opt) Cy5Bkgrd	Background2 value for Cy5
(opt) Cy3	RawIntensity1 value for Cy3
(opt) Cy5	RawIntensity2 value for Cy5

Automatic Gene Class naming based on Gene Name

Alternative Grid,Row,Column encoding scheme: NAME_GRC

      GRID-   grid#-Rrow#Ccol#

      GRID-   8-R12C11

Field	Description
(opt) Location	alternate spot identifier. E.g., Affymetrix 'probe_set', or Incyte 'IncyteID', etc. This may be numeric or alphanumeric
(opt) Clone ID	I.M.A.G.E. consortium database clone ID. It may have a "IMAGE:" or "ATCC:" prefix
(opt) Unigene cluster ID	NCBI UniGene database ID
(opt) dbEST3'	NCBI dbEST database
(opt) dbEST5'	NCBI dbEST database
(opt) GenBankId	NCBI GenBank database
(opt) GenBankId3'	NCBI GenBank database
(opt) GenBankId5'	NCBI GenBank database
(opt) RefSeqID	NCBI RefSeq database
(opt) LocusID	NCBI LocusLink database
(opt) OMIMID	NCBI OMIM database
(opt) SwissProtID	Swiss-Prot database

Parameter	Value	DataType	Comments
(opt) GenomicMenu1	GenBank	String	Name of the database. This will appear in the View menu
(opt) GenomicURL1	http://www.ncbi.nlm.nih.gov/htbin-post/Entrez/query?db=2&form=1&term=	String	URL to which one adds the 'GenomicIDreq' value
(opt) GenomicURLepilogue1		String	epilogue of the URL if any
(opt) GenomicIDreq1	GBID	String	Name of the GenomicID required and that is specified in the GIPO file as one of its fields
(opt) GenomicMenu2	UniGene	String	Name of the database. This will appear in the View menu
(opt) GenomicURL2	http://www.ncbi.nlm.nih.gov/UniGene/query.cgi?ORG=Mm&CID=	String	URL to which one adds the 'GenomicIDreq' value
(opt) GenomicURLepilogue2		String	epilogue of the URL if any
(opt) GenomicIDreq2	UID	String	Name of the GenomicID required and that is specified in the GIPO file as one of its fields