Extremely excited today to reveal the first of two great works (magnus opera?), just posted on bioarxiv, applying regulatory network analysis techniques to PDAC expression data to dissect the underlying biology of the disease. Tweetorial time! 1/

Eight years ago I launched an enduring collaboration with Andrea Califano @ColumbiaCancer to apply his regulatory network analysis techniques to #pancreaticcancer. So many amazing scientists have contributed to this work: @paslaise @hc_maurer @AlvaroCurielGa1 @ElyadaEla 2/

What is “regulatory network analysis”? At it's heart, it's a way of extracting more useful information from expression profiles. A fundamental flaw of differential gene expression (DGE) analysis is the assumption that each gene is independent. Biologist KNOW this is not true! 3/

DGE treats all ~25K detectable genes as SEPARATE variables, performs 25K T-tests, and then slaps on a multiple hypothesis correction to make the statisticians less dyspeptic. There is no consideration of the relationships between genes! 4/

Yet we KNOW genes are co-regulated in SETS by transcription factors and other REGULATORY FACTORS. Biologist: “p21 is a target of p53”. DGE: “Shhhh”. These relationships are completely ignored by expression analysis. THIS IS A TRAVESTY! 5/