GEO Features Analysis

BENEFIT ANALYSIS OF THE GENE EXPRESSION OMNIBUS (GEO)

Architecture

Datasets organized by Series, Samples and Platforms—advantageous for sorting data.

Data can be browsed if one does not want to perform search

Datasets encouraged to be MIAME compliant to standardize on public data

Easy to use interface with Java script for help box (help box changes when cursor is moved on different links)

.CEL files clearly labeled and downloadable directly off results page of search

Search

Search functions include limits (Boolean Operators e.g. AND, OR, NOT), history, clipboard, and help

Search available by platform, series or sample, as well as organism

EVS (Enterprise Vocabulary Service) integrated search

DataSets

Each dataset given a unique accession number for easy search access.

Entrez GEO search works in conjunction with other NCI databases including: GenBank, PubMed, Gene, UniGene, OMIM, Homologene, Taxonomy, SAGEMap, and MapViewer.

“Hold until published” feature available for private data

Login access can be given to reviewers for private data

DataSoft files available for download has table with each patient and expression values for each probe set.

Clearly lists array platform, PubMed ID organism studied

Web-based Analysis

Cluster heat maps: Hierarchical and K-means clustering algorithms provided. Regions of interest can be selected, enlarged, downloaded, plotted as line charts, or linked directly to Entrez GEO-Profiles.

Query subset A vs. B: Helps in the identification of genes that display differences in expression level between two sets of samples in a DataSet, calculated using t-tests or fold difference. Genes that meeting defined significance criteria displayed in Entrez GEO-Profiles.

Subset effects: Retrieves all profiles flagged as having significant effects with respect to a specific experimental variable, e.g. 'age' or 'strain'.

Value distribution: plots average gene expression for patients for comparison and outlier detection

Deposit formats supported

SOFTtext file

Multiple platform, sample and series data may be submitted as concatenated SOFT-formatted records.

Suitable when data already in a database, or if many samples to submit.

SOFT (Simple Omnibus Format in Text) is designed for rapid batch submission (and download) of data

SOFTmatrix (spreadsheet or text)

Multiple sample and series data may be submitted as a SOFT-formatted data matrix.

Suitable when data multiple hybridizations contained in one spreadsheet, e.g. multi-chip Affymetrix Pivot files.

Less suitable if sample data tables have many columns, e.g. GenePix data

MINiML

Multiple platform, sample and series data may be submitted as MINiML XML files.

MINiML (MIAME Notation in Markup Language) is data exchange format for microarray gene expression data, and other types of high-throughput molecular abundance data.

MAGE-ML

Can accept MAGE-ML formatted data, but MAGE-ML data can be structured in different ways, so we should review format and content of files to determine they are valid XML documents containing all minimum info. for submission.

• Processing times for MAGE-ML submission can be substantially longer than for our other deposit routes

Our database

Should be able to search data by genes, and biomarkers (whether they are up-regulated or down-regulated) for cross-study comparisons.

Search should be integrated with EVS

List co-expressed genes and locations on chip

No external data: .CEL files and data should be on FTP for easy access