Fig. 3From: SNooPer: a machine learning-based method for somatic variant identification from low-pass next-generation sequencingTraining assessment of Model 1A, 1B and 1C. Data used to construct these curves were obtained from SNooPer’s RF training phase using Dataset 2 as a validation set and a subset of Dataset 1 as training set. Dark cyan, blue and light blue represent SNooPer's Model 1A, 1B and 1C, respectively and AUCs are shown for each model. a ROC curves. Solid, dashed and dotted lines represent RF, C4.5 (J48) and SimpleCart algorithms respectively. TPR stands for True Positive Rate and FPR for False Positive Rate. b PR curves. c Cohen's Kappa coefficientBack to article page