Extended Data Fig. 2: The OncoSigRF and OncoSigNB algorithms produce highly similar KRAS LUAD SigMaps.
From: Oncoprotein-specific molecular interaction maps (SigMaps) for cancer network analyses
a, Comparison of ROC curves (FPR ≤ 0.05) for LUAD-specific KRAS SigMaps predicted by OncoSigNB (green and blue curves) and OncoSigRF (orange and red curve) trained on the Ingenuity PGSS and the MSigDB PGSS, respectively. b, Gene Set Enrichment Analysis (GSEA) of the top 100 OncoSigNB LUAD-specific KRAS SigMap predictions at the top of the OncoSigRF LUAD-specific KRAS SigMap predictions. Ranking is based on OncoSigRF score (SRF). Both the OncoSigNB predictions tested in the knockdown experiments (red lines) and the remaining top 100 OncoSigNB predictions (blue lines) are highly enriched at the top of the OncoSigRF predictions (p = 5.6 ×10−8 and p = 1.7 ×10-19, respectively).
