Inference of Personalized Drug Targets via Network Propagation
summary
Method simulates in silico knockouts on the HIPPY PPI network using random-walk with restart from patient mutations, repeats 1,000 random healthy propagations to build baseline distributions, then scores each node by how often its knockout pushes differentially expressed genes toward the healthy distribution to produce a back-to-healthy score. Applied to 174 AML patients from COSMIC, ranked targets are enriched for DrugBank entries and COSMIC causal genes, and FLT3-ITD subgroup predictions match known PI3K/MEK/PIM sensitivity trends.
AI-written summary · about