Performance of phenotype scores in predicting HT- and T2D-associated disease causal genes. ROCs for HT (A) and T2D (B). Replicated GAD genes or 'Intersection' genes shown in Fig. 3A were used as GSPs, and all the other genes were treated as GSNs to score all human genes using MGI or OMIM-phenotype-based predictors. For a given phenotype predictor, each gene in the genome was rank-ordered by the phenotype scores. Then at various rank cutoffs, the coverage, or the percentage of the total gold standard positives (GSPs) recovered, and the false positive prediction rate (1-specificity = FP/(FP+TN)) were calculated and plotted as the y- and x-axis variables on an ROC. FPs (false positives) are positive predictions which belong to gold standard negatives (GSNs). TNs (true negatives) are negative predictions which belong to GSNs. For control ROCs, the GSPs were replaced by the same number of randomly selected genes from the total 16023 genes we probed in this study.