Skip to main content
Fig. 4 | Journal of Cheminformatics

Fig. 4

From: A multi-label approach to target prediction taking ligand promiscuity into account

Fig. 4

MMM and SMMM target prediction performance on test sets. (a) Bar plots of the recall and precision values shown in columns 2 and 3 in Table 3, the performances of the MMM and SMM models for 16,344 single-label ChEMBL17 test compounds covering 308 target proteins. (b) Bar plots depict the recall and precision values (Columns 2 and 3 in Table 4) that illustrate the performance of both models for 4,403 single-label ChEMBL17 test compounds covering 63 7TM1 target proteins. (c) Bar plots represent the recall and precision values (Columns 4 and 5 in Table 4) returned by the two classification models for 2,887 single-label ChEMBL17 test compounds covering 89 Kinase target proteins. (d) Bar plots denote the recall and precision values (Columns 6 and 7 in Table 4), the target prediction performance of the models for 1,927 single-label ChEMBL17 test compounds covering 31 Protease target proteins

Back to article page