Msigdb v2 release notes

Details on how the gene set databases were generated is provided below:

C1 (Positional gene sets)

Cytogenetic locations were parsed from hugo (October 2006) and Unigene(build 197). When there were conflicts, the Unigene entry was used.

C2 (Curated gene sets)

C2 sets were curated from several sources including:

Online pathway databases: Several online resources provide catalogs of well studied metabolic and signaling pathways as well as functional categories of genes. We downloaded gene sets from 12 such databases into our system.

<tbody> </tbody>

Name	URL/Reference
BioCarta	http://www.biocarta.com
Signaling pathway database	http://www.grt.kyushu-u.ac.jp/spad/menu.html
Human cancer genome anatomy consortium	http://cgap.nci.nih.gov/
NetAffx	http://www.affymetrix.com/index.affx

Biomedical literature: Over the past few years, microarray studies have identified signatures of several important biological and clinical states (e.g. cancer metastasis, stem cell characteristics, drug resistance). These gene sets are valuable biological results. Unfortunately, because gene sets are typically published as tables in a paper, the important biological findings they represent are not easily accessible to computational tools. Our first goal was to convert published gene sets into an electronic form. Towards this we compiled a list of microarray articles with published gene expression signatures. From each article, we extracted one or more gene set from tables in the main text or supplementary information. Notably, our focus was on capturing the identity (e.g. gene symbol, GenBank accession) of all members in a gene set rather than on relationships between individual genes. Currently the process of curating a gene set from the literature is largely manual. In this report we include a collection of 1181 gene sets curated in this manner from 343 distinct PubMed accessions.

C3 (sequence motif gene sets)

We compiled gene sets on the basis of  shared regulatory motifs from a recently published comparative analysis of the Human, Mouse, Rat and Dog genomes (Xie, Lu et al. 2005). This database consists of 837 motifs sets including 222 microRNA target gene sets.

C4 (computed gene sets)

We mined 4 expression compendia datasets for correlated gene sets by searching for neighbors (i.e. genes with similar expression profiles across a compendium) of  380 cancer associated genes (Brentani, Caballero et al. 2003). Neighborhoods with <25 genes at a Pearson correlation threshold of 0.8 were omitted yielding 427 sets. This category of the database is identical to that previously reported in  (Subramanian, Tamayo et al. 2005).