Network Analysis of Cancer Mutations
summary
Presents an algorithmic fix to HotNet that uses asymmetric heat diffusion / lazy random-walk propagation on interaction graphs, followed by thresholding and strongly connected components to avoid TP53-driven artifacts. Runs on a pan-cancer set of ~3,000 samples, produces a consensus of 16 consensus subnetworks, highlights cohesin (~7% of samples) and condensin (~4%), and reports an FDR-controlled list of modules. Shows the tool finds rare, network-supported drivers that gene-by-gene tests miss.
AI-written summary · about