The influence of negative training set size on machine learning-based virtual screening

Table 1 The improvement of evaluating parameters calculated for targets from the initial tests

ML/Fingerprint		5-HT_1A			HIV PR			Metalloproteinase
		R	P	MCC	R	P	MCC	R	P	MCC
SMO	CDK FP	−0.19	0.71	0.54	−0.06	0.80	0.61	−0.18	0.89	0.72
SMO	MACCS	−0.34	0.20	0.23	−0.18	0.43	0.43	−0.35	0.58	0.47
NB	CDK FP	−0.07	0.03	0.05	−0.01	0.07	0.07	−0.05	0.05	0.06
NB	MACCS	−0.04	0.01	0.04	−0.04	0.03	0.04	−0.06	0.00	0.01
Ibk	CDK FP	−0.09	0.18	0.30	−0.03	0.42	0.46	−0.09	0.22	0.33
Ibk	MACCS	−0.1	0.08	0.17	−0.06	0.13	0.26	−0.11	0.13	0.25
J48	CDK FP	−0.22	0.09	0.16	−0.20	0.11	0.16	−0.17	0.11	0.20
J48	MACCS	−0.22	0.07	0.12	−0.16	0.11	0.18	−0.21	0.15	0.22
RF	CDK FP	−0.34	0.64	0.56	−0.10	0.66	0.60	−0.28	0.74	0.60
RF	MACCS	−0.20	0.22	0.31	−0.13	0.30	0.37	−0.19	0.39	0.43

The table shows the changes in given performance parameters for a particular ML method obtained between experiments with the lowest and the highest ratio of negative to positive training examples.

ISSN: 1758-2946