Table 1.

Performance of the EpiSearch method for six independent test cases. Results are shown for the highest scoring patch, for the gp120-mAb 17b complex we also include the data for the second scoring patch. In all six cases the highest scoring patches overlap with the experimentally known epitopes as shown in the number of correctly predicted residues (column 6). This number is then compared to the number of all residues in the epitope sites (column 4), given as coverage, TP/(TP = FN), in column 7, and also compared to the number of all predicted residues in that patch, given as overlap ratio, TP/(TP = FP), in column 8. TP = correctly predicted amino acids, FN = amino acids not predicted and FP = amino acids predicted incorrectly on the protein surface.

Name of antibody	Number of input peptides	Highest patch score	Number of interface residues*	Number of residues in a patch	Number of correctly predicted residues	Extent of coverage	Overlap ratio	CPU Time (second)
Trastuzumab	5	0.91	14	27	9	0.64	0.33	62
mAb Bo2C11	27	0.93	14	20	7	0.50	0.35	43
mAb b12	19	0.94	18	26	8	0.44	0.31	46
mAb 17b†1	11	0.78	11	27	2	0.18	0.07	29
mAb 17b†2	11	0.68	11	19	7	0.64	0.37	29
mAb 13b5	14	0.81	12	21	12	1.00	0.57	21
80R	18	0.84	16	22	6	0.38	0.27	24

^†Both first¹ and second² high scoring patches are shown.

^*Interface residues are calculated if the change is ASA more than 10Ų in complex formation.