Visualisation of variable binding pockets on protein surfaces by probabilistic analysis of related structure sets

BMC Bioinformatics

Table 2 Ensemble analysis of pockets at protein-protein interfaces

Complex	Protein-protein interface size (atoms)	Method	Percentage of interface identified as pocket
			*apo* structure	Provar Score > 0.75	Provar score > 0.25
Bcl-X_L		PASS	28	16	42
+	74	LIGSITE	66	39	83
Bak		fpocket	41	20	50
MDM2		PASS	27	11	45
+	193	LIGSITE	41	30	74
p53		fpocket	31	8	46
XiapBir3		PASS	30	9	55
+	128	LIGSITE	42	35	83
Caspase 9		fpocket	19	13	59
XiapBir3		PASS	17	0	43
+	42	LIGSITE	21	17	81
SMAD		fpocket	21	0	65
ZipA		PASS	54	13	65
+	63	LIGSITE	44	19	57
FtsZ		fpocket	10	2	19
HPVE2		PASS	42	22	51
+	109	LIGSITE	44	22	58
HPVE1		fpocket	17	3	32
IL2		PASS	30	5	55
+	103	LIGSITE	20	17	57
IL2-R		fpocket	4	1	19
Integrase		PASS	42	42	64
+	33	LIGSITE	45	18	48
LEDGF		fpocket	52	3	51
TNFa		PASS	23	9	44
trimer	189	LIGSITE	34	18	63
interface		fpocket	15	3	31
TNFR1a		PASS	24	5	34
+	41	LIGSITE	46	10	71
TNFb		fpocket	0	0	34
MDM4		PASS	55	24	48
+	71	LIGSITE	38	32	52
p53		fpocket	44	0	47
		PASS	34	14	49.5
Mean Values	95	LIGSITE	40.5	23.5	66
		fpocket	23	5	41

Provar analysis of a tCONCOORD generated ensemble of 250 structures of the apo form of each of the 11 first-named proteins in the complex. The proportion of protein interface atoms found to be persistently or variably lining pockets is, on average, significantly greater than for the protein as a whole. However, this enhanced variability is small and only found to be significant for 14/33 individual analyses. (Italic number in final column indicates the cases in which the proportion is notsignificantly greater at the p < 0.05 level).

ISSN: 1471-2105