Implementation of multiple-instance learning in drug activity prediction

BMC Bioinformatics

Table 8 Validations on the prediction of bioactive conformers

IDa	Nameb	PDB IDc	Contributiond	Ranke	n _i ^f
23	AR	1Q5K	2.792	3	117
37	Benzoimidazole-1	2O5K	0	N.A.^g	138
50	Jonjon-1	2OW3	2.827	6	38
59	LM-4	1Q3W	0.858	1	2
60	LM-5	1UV5	11.941	1	3
77	LM-29	1Q41	8.576	2	7
97	Maleimide	1R0E	0	N.A. ^g	121
98	OxaD-0	3F7Z	10.629	1	53
99	OxaD-00	3GB2	4.637	2	9
153	Pyzo-11	3L1S	10.371	1	11
198	RM-0	1Q4L	5.568	2	25
199	Staurosporine	1Q3D	22.359	1	5

^a Molecule index in the data set; ^b molecular name in the data set; ^c Protein Data Bank index for the protein structure from which the experimental conformer was extracted; ^d contribution $f (C_{i j^{*}})$ calculated using equation 6; ^e the rank in the set of contributions; ^f the number of conformers for each molecule; ^g the rank cannot be determined and the conformer was predicted to be irrelevant to classification based on the MILES method.

ISSN: 1471-2105