Table 5.

Overall accuracies (ACC) and MCC for our two methods (MLR and NN, with and without splicing) and the general methods (PMut, PolyPhen-2, PON-P2, and SIFT) in the CAGI dataset.

(A) BRCA1
IARC 5-tier	MLR	MLR+AS	NN	NN+AS	PMut	PolyPhen-2	PON-P2	SIFT
ACC	0.326	0.347	0.201	0.222	0.028	0.208	0.014	0.049
MCC	−0.041	0.006	0.015	0.056	−0.002	0.031	0	0.021
Three Class	MLR	MLR+AS	NN	NN+AS	PMut	PolyPhen-2	PON-P2	SIFT
ACC	0.764	0.785	0.25	0.271	0.035	0.354	0.014	0.118
MCC	−0.237	0.354	−0.012	0.055	0.026	0.136	0	0.123

(B) BRCA2
IARC 5-tier	MLR	MLR+AS	NN	NN+AS	PMut	PolyPhen-2	PON-P2	SIFT
ACC	0.161	0.167	0.345	0.351	0.144	0.305	0.011	0.034
MCC	−0.109	−0.068	−0.017	−0.006	−0.029	0.078	0	0.017
Three Class	MLR	MLR+AS	NN	NN+AS	PMut	PolyPhen-2	PON-P2	SIFT
ACC	0.931	0.931	0.69	0.695	0.184	0.431	0.011	0.086
MCC	0.18	0.277	0.185	0.213	−0.013	0.125	0	0.022