Reference Manual++ - MAExplorer Microarray Exploratory Data Analysis **Legend (Click on figures to show higher resolution versions) *** DRAFT *** $Date: 2003/11/23 10:29 $ MicroArray Explorer-V0.96.34.01 "Millenium" edition Peter F. Lemkin Laboratory of Experimental and Computational Biology Center for Cancer Research, National Cancer Institute - Frederick, Frederick, MD 21702

MAExplorer home | Table of Contents | Overview | Introduction | Menu summary | PDF documents | Newsletters | Plugins | Quick start | Short tutorial | Advanced Tutorial | Glossary | Figures | Tables | Index | Help desk

SourceForge     http://maexplorer.sourceforge.net/   or LECB/NCI       http://www.lecb.ncifcrf.gov/MAExplorer

++ Note: This hypertext manual is divided into chapters and appendices Web pages. These may be printed individually from your Web browser by (1) clicking in the text window to be printed, and (2) using the "Print Frame" in Netscape or "Print" in Internet Explorer. Some of the chapters (eg. 2) have many images. The entire manual may be downloaded at one time with low resolution figures and is suitable for printing in the Web browser. You may also download a an Adobe acrobate PDF file version of the entire manual with the lower resolution figures (~5Mb). The Unix script for creating the full reference manual from the individual HTML pages is CreateMaeFullRefManual.do.

The MAExplorer is a Java-based bioinformatics exploratory data-analysis and data-mining program for analyzing sets of quantitative spotted cDNA or oligonucleotide microarray data (Lemkin et al., 2000) - (see (Schulze, 2001) for a review of microarray technology).

Prior to its release on SourceForge, MAExplorer was developed by Dr. Peter Lemkin (LECB/NCI-Frederick) with help from Gregory Thornwall (SAIC) and Jai Evans (DECA/CIT, NIH). It was initially created for analyzing ³³P labeled membrane array data from the mouse mammary tissue from Mammary Genome Anatomy Project (MGAP) http://mammary.nih.gov/ with the help of many researchers in the Laboratory of Genetics and Physiology, NIDDK under Dr. Lothar Hennighausen. Since the early work with MGAP it was extended to work with other types of cDNA and oligo arrays and various nucleotide labeling methods. These include spotted Cy3/Cy5 glass slides, spotted membranes, non-geometric chip data, and other chip supports with different geometries and numbers of duplicate spots/gene, clones as well as oligo chip data such as Affymetrix. A wizard tool called Cvt2Mae was developed to make it easier for other researchers to convert their data to the format required by MAExplorer. Cvt2Mae was developed by Peter Lemkin, Greg Thornwall and Bob Stephens (ABCC/SAIC). You may extend the set of builtin analysis methods by writing Java plugins called MAEPlugins.

This document describes the MAExplorer's functionality, provides tutorials and contains documentation for using it with various types arrays.

With this program, you may: 1) analyze expression of individual genes; 2) analyze expression of gene families and clusters; 3) compare expression patterns for multiple hybridized samples.

MAExplorer is written in Java and runs as a stand-alone application that you download to your computer. Although MAExplorer began out as a Java applet for use with with Web browsers for the MGAP Web database ( http://www.lecb.ncifcrf.gov/mae ), we have depricated its use as an applet because of many problems with running large Java applets in some Web browsers. Instead, we recommend downloading MAExplorer which includes the public MGAP array data as a demonstration data set. Then run MAExplorer on this data after you have installed it on your computer.

Notation: MAExplorer uses the notation that the sample probe total mRNA is labeled and then hybridized against the known cDNA targets tethered to the microarray. Because of this notation, we refer to a hybridized sample as a HP. An alternative notation that reverses these terms is also commonly used (see "Chipping Forecast", Nature Genetics supplement, Jan, 1999, pg 1). Also, because arrays may be constructed from either spotted clones or oligonucleotides, we refer to hybridized chip DNA from any of these sources genericlly as "genes".

Throughout this document we use the abbreviations HP for hybridized sample, GC for gene class. These and other terms are explained in the Glossary and Index . There are a number of figures and tables illustrating various features of MAExplorer throughout this manual. Figures are presented at low-resolution. By clicking on the lower-resolution figure, the high-resolution versions can be viewed.

NOTES: because MAExplorer is under development, there may be occasional problems with some of its functionality. There may also be some problems (mostly bad HTML links) with migrating from LECB/NCI to the SourceForge Web site. Some operations that are under development are labeled with "[Future]" in this manual. We welcome your suggestions for improvements as well as letting us know about problems that you encounter. Occasionally the manual or the figures in the manual may not be quite in phase with the software. Please notify us of problems or suggestions by E-mail so we can try to fix or implement them. If you are a bioinformatics developer and would be interested on working with the MAExplorer project, consider joining the MAExplorer development team on SourceForge.net.

**Icon Legend

MAExplorer - Overview

• MAExplorer helps perform computer data-mining of multiple samples hybridized with microarrays. Data mining is the process of attempting to find relevant patterns of information from large sets of data. MAExplorer enables the investigator to:
  1. organize the hybridized sample data by experimental condition, (including: disease state, dose response, developmental stage, strain, time course, knock-in/-out, shock treatment, etc.) so an investigator can design data mining experiments relevant to those conditions or a subset of conditions from a particular database.
  2. compare gene expression patterns between sets of different hybridized samples (denoted HP-X and HP-Y 'sets') for comparing mean changes between replicate sets of samples. An ordered expression profile list HP-E of samples is used for finding similar expression patterns across genes for a sequence of samples such as from the cell cycle, developmental stage, or conditions.
  3. use data mining techniques of graphical direct-manipulation (which requires the real-time response of local computation), statistical, clustering and spreadsheet techniques, and connectivity to other Internet genomic databases to get additional information on individual genes. The latter leverages the maintenance resources of other groups to allow transparent access to that data.
  4. explore, compare, and record analyses between researchers in their own group and for sharing with other investigators (i.e. groupware).

• Spots in microarray images are quantified into tab-delimited data files using programs such as generated by Axon's GenePix^(TM), Scanalyze, Molecular Dynamic's ImageQuant-NT^(TM), Research Genetics' Pathways^(TM), and other systems. This data is transformed using the Cvt2Mae wizard data conversion tool to quantification data files, a print-file (Gene In Plate Order table or GIPO), a list of DB samples file, and a MAExplorer configuration file. These may be copied to your local computer file system (in stand-alone operation) or a MAExplorer-compatible Web database server where they are loaded on demand by MAExplorer. The file formats schema used by MAExplorer is discussed in Appendix C). The data conversion tool Cvt2Mae (Appendix C.6) helps convert user's data sets to MAExplorer format.

• Upon starting, MAExplorer uses a ".mae" startup file to specify the subset of samples to be used from the database, and initial parameters to use. It then loads a configuration file which describes additional files including the gene-in-plate-order table which maps spot position to clone ID and other genomic information, and a samples database file containing a list of the names of the quantified spot data files for the hybridized samples being analyzed. Later, you may request additional sample files or data from the local file system or Web database server when requested by the user. The .mae file format is discussed in Appendix D. Users may save data mining sessions to create new .mae startup files. These may be used at a later for continuing their sessions in the future.

• The investigator interacts directly with the system by selecting entries from menus (Section 2), selecting data by clicking on spots in the microarray image, selecting points in graphic plots or cells in spreadsheets, manipulating threshold sliders, or typing in gene names, clone IDs, GenBank IDs, UniGene IDs, LocusLink IDs, etc. Data reports may be exported to Excel spreadsheets or used dynamically to access other genomic Web databases. With the stand-alone version, you may save the full resolution plots in GIF files and report tables in tab-delimited text files.

Recommended Hardware

• A computer with at least 500Mhz CPU speed (Intel). For other CPUs such as the PowerPC (Macintosh), Sparc (Sun), etc., it should have a corresponding capability (for more powerful CPU chips, a lower CPU speed may be fine). For large data sets (order of 100 or more) having a large number of samples with many spots, a much faster system with much more memory is desirable.
• At least 128Mbytes of memory. Although it will work with less memory, we don't recommend it as it is underpowered. For large data sets, more memory (eg. 256Mb or more) is desirable.
• Adequate disk space for the data sets required. The MAExplorer distribution itself, including the MGAP demonstration database and a Java Virtual Machine, is on the order of 24 Mbytes. This Reference Manual with both low and high resolution figures is on the order of 11 Mbytes.
• It requires at least a 1024x768 pixel resolution 256-color monitor. However, we find that because of the multiple plots created during a session, it is much easier to use with a screen resolution of 1280x1024 pixels. It is very difficult to use with an 800x600 resolution system and we don't recommend it.
• The R extensions are not available with MacOS 8/9. Using the R extensions requires more memory - at least 256Mbytes with a faster processor is recommended.

Addition of user defined analysis methods using Java Plugins