Machine learning to predict metabolic drug interactions related to cytochrome P450 isozymes

Table 3 The accuracy values of the substrate models based on the 10 randomly-generated negative datasets

	Des	1A2_Sub	2C9_Sub	2C19_Sub	2D6_Sub	3A4_Sub
1	2D	0.70 ± 0.049	0.72 ± 0.042	0.74 ± 0.045	0.78 ± 0.031	0.75 ± 0.025
2	2D	0.69 ± 0.041	0.74 ± 0.041	0.73 ± 0.050	0.78 ± 0.028	0.77 ± 0.024
3	2D	0.71 ± 0.045	0.73 ± 0.041	0.75 ± 0.045	0.78 ± 0.031	0.76 ± 0.023
4	2D	0.69 ± 0.045	0.72 ± 0.048	0.75 ± 0.046	0.78 ± 0.031	0.74 ± 0.020
5	2D	0.72 ± 0.047	0.76 ± 0.035	0.74 ± 0.047	0.77 ± 0.033	0.76 ± 0.019
6	2D	0.72 ± 0.042	0.71 ± 0.045	0.74 ± 0.049	0.78 ± 0.035	0.75 ± 0.023
7	2D	0.71 ± 0.036	0.74 ± 0.040	0.75 ± 0.045	0.78 ± 0.031	0.75 ± 0.024
8	2D	0.72 ± 0.050	0.73 ± 0.046	0.74 ± 0.051	0.77 ± 0.033	0.75 ± 0.020
9	2D	0.72 ± 0.041	0.73 ± 0.042	0.77 ± 0.040	0.79 ± 0.032	0.75 ± 0.022
10	2D	0.71 ± 0.046	0.72 ± 0.045	0.73 ± 0.043	0.78 ± 0.034	0.74 ± 0.020

Bold value represents the accuracy of the best of the 10 randomly generated negative sets

ISSN: 1758-2946