Fig. 2: An ensemble random forest classifier to predict SV gene pairs.
From: Synthetic viability induces resistance to immune checkpoint inhibitors in cancer cells
a The weights of 14 features from random forest classifier. b ROC curve for ensemble classifier. c, d The AUC and precision for ensemble classifier’s performance in predicting published resistant genes. e The regression coefficients of SV gene pairs were estimated by multiple linear regression model (FDR-adjusted P < 0.05). Significant differential dependency scores of EGFR in cancer cell lines where FGFR1 (f) or PDGFRA (g) was gain of function versus where FGFR1 or PDGFRA was wild type. P values were calculated from the multiple linear regression model and FDR-adjusted P < 0.05 was considered statistically significant.
