Table 1.

Sensitivity, specificity, type II error rate, type I error rate, missing values (NAs), total accuracy and area under ROC curve estimation for PolyPhen-2, SIFT and CAROL using an optimal threshold of 0.57, 0.96 and 0.98, respectively

	PolyPhen-2	SIFT	CAROL
Sensitivity	0.800	0.819	0.830
Specificity	0.705	0.727	0.727
Type II error	0.200	0.181	0.170
Type I error	0.295	0.273	0.273
NAs	0.116–0.125	0.004–0.177	0.001–0.072
Total accuracy	0.721	0.744	0.745
ROC area	0.836	0.821	0.852

Numbers in bold denote where CAROL performs better than the other 2 annotation tools.

Sensitivity equals the number of true positives divided by the number of true positives plus false negatives; Specificity equals the number of true negatives divided by the number of true negatives plus false positives; Type II error equals 1 – sensitivity; Type II error equals 1 – specificity; Total accuracy equals the number of true positives plus negatives divided by the number of total positives plus negatives; ROC area equals the value of the Wilcoxon-Mann-Whitney test statistic [as in ref. 32].