Table 1.

Classification Performance of SVM Models Developed Based on Dataset Train1 with Various Descriptor Sets

	MOE	Schrödinger	ISIDA fragments
10-fold cross-validation (61 actives and 73 inactives)
Parameters	C = 5 γ = 0.1	C = 20 γ = 0.1	C = 40 γ = 0.1
Accuracy	95.44%	95.44%	91.65%
Sensitivity	96.72%	95.08%	88.52%
Specificity	94.52%	95.89%	94.52%
Prediction on external test set Test1 (41 actives and 49 inactives)
Accuracy	94.44%	90.00%	91.11%
Sensitivity	95.12%	87.80%	87.80%
Specificity	93.88%	91.84%	93.88%
Prediction on external test set DBB (284 weak actives)
Accuracy	18.31%	17.61%	22.18%
Prediction on external test set DBC (847 inactives)
Accuracy	99.65%	99.29%	96.81%