highly significantP = 1.5 × 10 −149
Although the overall results were highly significant for all methods, P values from the tests were the most significant for ReMM ( P = 1.5 × 10 −149 ), FATHMM noncoding ( P = 2.9 × 10 −121 ) and Eigen ( P = 2.5 × 10 −114 ), while P values for fitCons ( P = 6.0 × 10 −11 ) and GWAVA unmatched ( P = 7.3 × 10 −13 ) were the least significant, consistent with the AUC results.