MASS MINING: A CROWDSOURCING APPROACH FOR META-ANALYZING GENE EXPRESSION SIGNATURES OF AUTOIMMUNITY USING LARGE-SCALE PUBLIC DATA SETS

Abstract The volume and diversity of large-scale biological data available in the public domain continues to grow. This data has the potential to be reused to answer questions beyond those envisioned when the data was generated; however, few immunologists have sufficient bioinformatics expertise to do so. We used OMiCC, a free online platform that enables programming-free meta-analysis of public gene expression data and facilitates “crowdsharing” the work of annotating and constructing data compendia. We organized an “OMiCC Jamboree” to evaluate if biologists without bioinformatics training could use OMiCC to identify and annotate public gene expression datasets and design proper disease versus control comparisons for meta-analysis. Twenty-nine volunteer NIH biologists gathered to search and annotate public microarray data of human autoimmune conditions and the corresponding mouse models. Meta-analyses across studies explored 1) gene expression signatures for each disease, 2) pan-disease signatures, and 3) cross-species signatures. A large number of differentially expressed genes and enriched pathways were identified for each disease, with substantial overlap among diseases both within and between species, including pan-disease and pan-species signatures such as those associated with interferon. Supported by the Intramural Research Programs of NIAID and CIT, NIH.