Deep learning-based classification model for GPR151 activator activity prediction

BMC Bioinformatics

Table 8 Comparison of different feature selection algorithms on test set performance. The results of traditional methods in the table are the best with five classifiers, and the bold marks the best in the group

Feature	Algorithm	Accuracy	Precisionn	Recall	F1
RDKFP	PCA	0.8958	0.8291	0.5391	0.6534
	LDA	0.8583	0.6089	0.6214	0.6151
	DTA	0.8576	0.6626	0.4444	0.532
	CNN	0.9003	0.7806	0.6296	0.697
MorganFP	PCA	0.8988	0.7647	0.642	0.698
	LDA	0.8726	0.654	0.6379	0.6458
	DTA	0.8816	0.7707	0.4979	0.605
	LSTM	0.8981	0.7082	0.749	0.728
Mol2vec	PCA	0.8816	0.7157	0.5802	0.6409
	LDA	0.8876	0.736	0.5967	0.6591
	DTA	0.8928	0.7525	0.6132	0.6757
	CNN	0.9153	0.7928	0.7243	0.757

ISSN: 1471-2105