Table 5.

Comparison of the predictive performance of Pep-3D-Search with different distance parameters (AHA).

PDB ID	1jrh	1bj1	1g9m	1e6j	1n8z	1iqds	1yy9	2adf	1avz	1hx1	Average
AHA (distance threshold = 3.7)
TP/PE	17/20	7/16	12/44	11/34	17/36*	6/32	6/25*	9/33	13/36	8/32
MCC	0.375	0.1243	0.1716	0.2286	0.1565	0.0733	0.0595	0.1502	0.2855	0.0351	0.1659
Sensitivity	0.85	0.4375	0.2727	0.3235	0.4722	0.1875	0.24	0.2727	0.3611	0.25	0.3667
Precision	0.8095	0.3684	0.8	1.0	0.85	0.375	0.4	0.6	0.8125	0.3333	0.6349
AHA (distance threshold = 4)
TP/PE	16/20	7/10	13/42	8/30	15/30*	5/39	6/23	10/37	12/35	10/34
MCC	0.3491	0.1528	0.1887	0.1584	0.1379	0.0283	0.0598	0.1695	0.2525	0.0858	0.1583
Sensitivity	0.8	0.7	0.3095	0.2667	0.5	0.1282	0.2609	0.2703	0.3429	0.2941	0.3873
Precision	0.7619	0.3684	0.8667	0.7273	0.75	0.3125	0.4	0.6667	0.75	0.4167	0.6021
AHA (distance threshold = 4.3)
TP/PE	16/18	8/11	14/42	8/36	17/30*	9/37	4/16*	11/37	10/33	4/26
MCC	0.3537	0.1773	0.2052	0.1558	0.1571	0.1431	0.0394	0.1923	0.1869	-0.0514	0.1559
Sensitivity	0.8889	0.7273	0.3333	0.2222	0.5667	0.2432	0.25	0.2973	0.303	0.1538	0.3986
Precision	0.7619	0.4211	0.9333	0.7273	0.85	0.5625	0.2667	0.7333	0.625	0.1667	0.6048
AHA (distance threshold = 4.6)
TP/PE	19/36	9/24	9/44	4/41	18/38	0/34	8/25*	8/36*	5/37	6/27
MCC	0.3989	0.1483	0.1206	0.0495	0.1663	-0.1023	0.0813	0.1241	-0.0357	0.0051	0.0956
Sensitivity	0.5278	0.375	0.2045	0.0976	0.4737	0.0	0.32	0.2222	0.1351	0.2222	0.2578
Precision	0.9048	0.4737	0.6	0.3636	0.9	0.0	0.5333	0.5333	0.3125	0.25	0.4871

TP: number of true positives; PE: number of residues in the predicted epitope; TN: number of true negatives; FP: number of false positives; FN: number of false negatives; Matthews correlation coefficient $(\text{MCC})=\frac{(TP\cdot TN)-(FP\cdot FN)}{\sqrt{(TP+FP)(TP+TN)(TN+FP)(TN+FN)}}$; $\text{Sensitivity }(\text{Se})=\frac{TP}{TP+FN}$; $\text{Precision }(\text{Pr})=\frac{TP}{TP+FP}$.

*For the cases, the best prediction was found in the second-ranked candidate.