Meta-modules

Created By Denise Wolf dwolf
The file TCGA_Pancan12_metamodules22.csv contains the meta-module scores for the Pancan12 RNA-seq gene expression data (PanCan12.3602-corrected-v3). These 22 meta-modules were derived from an analysis of thousands of gene signatures, pathways, modules and metagene attractors, in an effort to collapse these signatures into a smaller set of non-redundant features. To derive these meta-modules, essentially 'consensus' groupings of highly correlated signatures or pathways, we first filtered most of the signatures for bimodal expression patterns, including only those with bimodal distributions in at least one cancer (bimodal index BI>1.1). We reasoned that bimodal signatures might be helpful for understanding cancer classifications that transcend tissue of origin, as they capture 'switch-like' heterogeneity within cancers. This filtering was performed for the MsigDB signatures and pathways, but not the metagene attractors, as the method by which the latter were produced can be thought of as implementing high-order significance filtering. The next step was to apply weighted gene correlation network analysis (WCGNA) to the filtered and metagene attractor signatures in order to reduce this still large set to a smaller non-redundant set of features. Doing so resulted in 22 'meta-modules' - non-redundant co-expressed modules of signatures/modules that capture the information in the larger, correlated set. These meta-modules have been named to reflect their composition (E.G., Proliferation and Immune-interferon). Also included is a ppt file that outlines our approach, with references to the R packages used for the analyses and a description of the meta-modules, including the number of composite signatures in each. Contributors to this suite of analyses include me (Denise Wolf), Josh Stuart, Chuck Perou, Katie Hoadley, Chris Feng, Dimitris Anastassiou, Weiyi Cheng, Tai-Hsien Ou Yang, Laura Van ?t Veer & TCGA pancan12 working group. If you'd like help with analyses of this type for any cancer or cancer grouping, please let me know as I'd like to be of service (Denise.Wolf@ucsf.edu).

syn1960752
syn1960760
syn1960753
syn1960751
syn1960647
syn2227907
syn2018908
syn1809694
syn1809703
syn1809704
syn1960763
syn1809698
syn1809701
syn1809699
syn1960758