Efficient conformational sampling and weak scoring in docking programs? Strategy of the wisdom of crowds

Table 1 Comparison of the efficacy of the programs

Programs: sampling methods and scoring functions	Conformational sampling			Scoring
Programs: sampling methods and scoring functions	1-Number of poses obtained	2-Number of correct poses	3-Number of targets with a correct pose	4-Number of targets with a correct pose ranked in the top 4	5-Number of targets with a correct pose ranked as the top 1
FlexX	2915	910	65
FlexX				57	52
Surflex	2899	1152	84
Surflex				72	56
Glide-SP	2393	624	79
Glidescore				75	65
Emodel				72	65
Glide-XP	392	210	74
Glidescore				73	68
Emodel				73	66
Gold	1447	330	77
PLP				74	64
Goldscore				73	60
USC based on docking results				87

The comparison is made for ligand conformational sampling (columns 1–3) and pose scoring (columns 4 and 5). From the requested 3000 poses per program (30 poses per target, for 100 targets), the number of obtained poses is given in the first column. Of these poses a certain number is correct, with RMSD < 2 Å from the crystal position, (column 2) and corresponding to a number of targets (column 3). The number of targets whose correct poses are ranked in the top 4 are given in column 4 and those whose correct poses are top-ranked are given in column 5. The number of targets with a correct pose obtained with the USC method, based on the docking results, is reported in the last line (see the “USC method” section below)

ISSN: 1758-2946