Table 3.

Prediction results for the training set with 6 combined terms

Complex	Unbound	No. of Surface Residues	No. of Interface Residues	Sensitivitya	Precisiona	Specificitya	AUC
1AHW_AB:C	1TFH_A	173	25	0.360	0.153	0.662	0.481
1BVK_DE:F	3LZT_	98	17	0.765	0.361	0.716	0.835
1DQJ_AB:C	3LZT_	98	20	0.300	0.182	0.654	0.534
1E6J_HL:P	1A43_	63	13	0.462	0.353	0.780	0.585
1JPS_HL:T	1TFH_B	155	25	0.360	0.170	0.662	0.517
1 MLC_AB:E	3LZT_	98	16	0.562	0.250	0.671	0.636
1VFB_AB:C	8LYZ_	107	18	0.833	0.385	0.730	0.833
1WEJ_HL:F	1HRC_	95	13	0.462	0.188	0.683	0.649
2VIS_AB:C	2VIU_A	247	20	0.900	0.281	0.797	0.901
1BJ1_HLJK:VWb	2VPF_GH	160	35	0.600	0.412	0.760	0.705
1FSK_BC:A	1BV1_	145	19	0.526	0.233	0.738	0.587
1I9R_HL:ABCb	1ALY_ABC	320	65	0.508	0.292	0.686	0.687
1IQD_AB:C	1D7P_M	127	17	0.765	0.361	0.791	0.848
1K4C_AB:C	1JVM_A	88	16	0.500	0.258	0.681	0.647
1KXQ_H:A	1PPI_	341	30	0.600	0.148	0.666	0.637
1NCA_HL:N	7NN9_	263	27	0.556	0.163	0.674	0.684
1NSN_HL:S	1KDC_	106	23	0.174	0.114	0.627	0.454
1QFW_HL:AB	1HRP_AB	170	17	0.235	0.071	0.660	0.484
1QFW_IM:AB	1HRP_AB	170	17	0.706	0.214	0.712	0.738
2JEL_HL:P	1POH_	68	18	0.167	0.158	0.680	0.498
1BGX_HL:T	1CMW_A	646	66	0.394	0.124	0.683	0.521
2HMI_CD:AB	1S6P_AB	810	14	0.429	0.024	0.697	0.518
Mean		207	24.1	50.7	0.222	0.7	0.635

^a Sensitivity, precision, and specificity were recorded when 55% of surface residues were predicted as interface residues by the single term. We chose the parameter (55%) so that the sensitivity was about 50% in the consensus prediction. ^bMultiple binding sites.