Fig. 2

PPV-Sensitivity and ROC curves for 14 transcription factors. In parentheses the number of know kinases interacting with each TF present in the “Golden Standard”. A pre-filtering step based on the Fold Change (FC) of the modulator was applied to remove kinases with a FC ≤ 1 (Material and Methods). Positive Predicted Value (PPV) or precision is computed as a fraction of TP/ (TP + FP). Sensitivity (or true positive rate, TPR) is computed as a fraction of TP/ (TP + FP). True Negative Rate (TNR) is coputed as 1 – Specificity with Specificity equal to TP/ (TP + FP). a The cumulative PPV-Sensitivity curve of DMI across the 14 transcription factor obtained by averaging the individual PPV-sensitivity curves of each TFs (Material and Methods); b The cumulative receiver operating characteristic (ROC) curve of DMI across the 14 transcription factor (Material and Methods); c PPV-sensitivity curve for each one of the 14 transcription factor in which we compared the performance of DMI with and without applying a significance threshold for the p-value (P < 0.05) after Benjamini-Hochberg correction; d ROC curve for each one of the 14 transcription factor in which we compared the performance of DMI applying a significance threshold for the p-value (P < 0.05) after Benjamini-Hochberg correction