Finding correlated genes on TCGA expression data.

We have published a new section in IntOGen in collaboration with Hautaniemi Lab, where you can do a gene correlation with the gene expression results of TCGA (The Cancer Genome Atlas) for a given subset of patients that share some specific clinical annotations. Then you will be able to browse your correlations next to global gene expression, SNP survival and DNA methylation analyses. The data analyzed so far includes 4 tumor types from TCGA: glioblastoma multiforme (GBM; 507 samples), ovarian serous adenocarcinoma (OV; 546 samples), breast invasive carcinoma (BRCA; 525 samples) and colon adenocarcinoma (COAD; 161 samples).

The analysis has been done using Anduril, a workflow framework developed at Hautaniemi lab. The overall goal in the analyses is to identify common genomic regions or transcripts that have survival effect in individual cancers. To this end we have identified in all these four cancers differentially expressed genes, genomic regions with copy number aberrations or differential methylation, single nucleotide polymorphisms (SNPs), genes and genomic regions with significant survival association with Kaplan-Meier method, and genes that have simultaneous copy number alteration and significant expression changes.

Since we created IntOGen we had the motivation to let the user do simple analyses over the data and browse their results next to our datasets. For this reason we are developing Onexus, analysis management system that integrates the IntOGen browser with Anduril. This allows to create websites where the final user can define new analysis, run them on the fly and browse the results next to precalculated datasets. Onexus is under development, we are planning a first public release by the first 2012 quarter, if you want to learn more about it visit the Onexus web site.

Today Kristian Ovaska, from Hautaniemi lab, is presenting this work at the TCGA symposium.