Ensemble Learning Prediction of Drug-Target Interactions Using GIST Descriptor Extracted from PSSM-Based Evolutionary Information

Table 6

5-fold cross-validation results were generated by using the proposed RF classifier and SVM classifier on the enzyme dataset.

Testing set	Accuracy (%)	Precision (%)	Sensitivity (%)	MCC (%)	AUC
PSSM+GIST+RF
1	88.89	89.70	87.86	80.24	0.9570
2	89.40	89.90	88.98	81.05	0.9450
3	88.63	88.87	88.10	79.85	0.9444
4	89.91	91.97	87.20	81.83	0.9538
5	89.15	91.88	86.13	80.62	0.9391
Average	89.20 ± 0.49	90.46 ± 1.39	87.65 ± 1.07	80.72 ± 0.76	0.9479 ± 0.0073
PSSM+GIST+SVM
1	81.11	82.85	78.46	69.32	0.8794
2	82.65	83.65	81.53	71.31	0.8895
3	82.22	83.94	79.31	70.71	0.8901
4	81.28	82.23	79.24	69.53	0.8820
5	81.88	84.02	79.19	70.29	0.8770
Average	81.83 ± 0.64	83.34 ± 0.78	79.54 ± 1.16	70.23 ± 0.83	0.8836 ± 0.0059