Scoring functions and enrichment: a case study on Hsp90

BMC Bioinformatics

Table 3 Receiver operating characteristic data obtained with the Thr184 tether

Scoring Functions	GOLD_rank					BestScore_rank
	% Actives					% Actives
	10	20	30	50	AUC	10	20	30	50	AUC
*GOLD*	37	61	78	92	.824	63	87	96	99	.919
*ChemScore*	33	52	70	90	.797	56	81	93	99	.898
*DOCK*	32	47	62	81	.753	59	84	92	100	.899
*PMF*	0	6	19	43	.466	3	4	9	53	.505
*BLEEP*	28	44	57	79	.718	45	73	87	98	.867
*Consensus*	39	53	69	92	.803	63	92	97	100	.930
*GOLD* (normalised)	43	60	73	88	.806	79	94	98	99	.954
*ChemScore* (normalised)	24	44	62	86	.751	52	76	88	96	.878
*DOCK* (normalised)	35	49	63	86	.769	90	96	97	100	.976
*PMF* (normalised)	0	2	14	41	.448	3	4	13	56	.521
*BLEEP* (normalised)	25	39	52	76	.699	69	83	90	96	.909
*Consensus* (normalised) (normalised)	38	57	71	89	.801	89	97	99	100	.974

Retrieval of actives with the Thr184 tether. Percentages of actives corresponding to the top 10%, 20%, 30% and 50% of the screened library. The normalised scores were obtained by dividing by the number of heavy atoms to the power of 1/3.

ISSN: 1471-2105