Predictive value of the seven evolutionary metrics for individual disease-causing CFTR mutations.

For each evolutionary metric, ROC analysis was used to evaluate the trade-off between sensitivity and specificity over a range of score thresholds, and the optimal threshold was taken as position of on the curve that minimizes the Euclidean distance from the perfect predictor (see Figure 7 and text). For these thresholds, evolutionary metrics are ranked according to the harmonic mean between the sensitivity and specificity of predicting individual CFTR residues as disease-causing or non-mutated.