Skip to main content
. Author manuscript; available in PMC: 2020 Sep 1.
Published in final edited form as: Hum Mutat. 2019 Jul 3;40(9):1593–1611. doi: 10.1002/humu.23802

Figure 4. Binary, cross-validated performance of the predictors.

Figure 4.

We represent the performance of our MLR and NN methods, as well as that of general predictors (CADD, PolyPhen-2, PMut, PON-P2 and SIFT), using four parameters: accuracy and MCC (radar plots (A), (B), (C) and (D)) and sensitivity and specificity (scatterplots, (E) and (F)). The methods labeled MLR-CAGI and NN-CAGI are those used to generate our CAGI predictions; for completeness, we give the performance of the other versions: MLR-psMSA (entropy and pssmnat values were obtained from psMSA-based parameters) and NN-oMSA (entropy and pssmnat values were obtained from oMSA-based parameters). In (E) and (F) points are colored according to the set in which sensitivity and specificity were estimated: blue and orange for the MLR and NN sets, respectively.