Exploiting protein flexibility to predict the location of allosteric sites

BMC Bioinformatics

Table 1 Prediction results on protein allosteric sites

Set	TP+FP	TP	FP	FN	Sensitivity	Specificity	Accuracy	PPV
Total	464	58	406	0	1.00	0.00	0.13	0.13
F	117	32	85	26	0.55	0.79	0.76	0.27
S	108	24	84	34	0.41	0.79	0.75	0.22
FS	36	15	21	43	0.26	0.95	0.86	0.42
c123	174	44	130	14	0.76	0.68	0.69	0.25
c123F	74	29	45	29	0.50	0.89	0.84	0.39
c123S	55	22	33	36	0.38	0.92	0.85	0.40
c123FS	30	14	16	44	0.24	0.96	0.87	0.47
c1	58	26	32	32	0.45	0.92	0.86	0.45
c1F	42	22	20	36	0.38	0.95	0.88	0.52
c1S	25	15	10	43	0.26	0.98	0.89	0.60
c1FS	20	13	7	45	0.22	0.98	0.89	0.65

Results on this table refer to the subset of 58 proteins for which LIGSITE^cswas able to predict a ligand-binding pocket in the position of the allosteric site. TP: true positive; TN: true negative; FP: false positive; FN: false negative; PPV: positive predictive value. Sensitivity: TP/(TP+FN); specificity: TN/(TN+FP); accuracy: (TP+TN)/(TP+FN+TN+FP); PPV or precision: TP/(TP+FP). The total number of pockets considered, predicted by LIGSITE^cs, is 464 (8 per protein). F corresponds to sets including a change in flexibility as selection criterion; S corresponds to sets including high structural conservation as selection criterion; c123 refers to sets considering only the three largest pockets predicted by LIGSITE^cs; c1 refers to sets considering only the largest predicted pocket.

ISSN: 1471-2105