Table 3.

Performance evaluation of the DTA prediction models on the KIBA dataset.

Methods	Proteins	Compounds	MSE ↓	CI ↑	${\mathit{r}}_{\mathit{m}}^2$↑	Pearson ↑
KronRLS	Smith–Waterman	Pubchem-Sim	0.411	0.782	0.342	-
SimBoost	Smith–Waterman	Pubchem-Sim	0.222	0.836	0.629	-
DeepDTA	CNN	CNN	0.194	0.863	0.673	-
WideDTA	CNN + PDM	CNN + LMCS	0.179	0.875	-	0.856
GraphDTA	CNN	GAT − GCN	0.139	0.891	-	-
MGraphDTA	MCNN	MGNN	0.128	0.902	0.801	-
WGNN-DTA	GCN	GCN	0.144	0.885	0.781	0.888
WGNN-DTA	GAT	GAT	0.130	0.898	0.791	0.899
DGraphDTA	GCN	GCN	0.126	0.904	0.786	0.903
MSGNN-DTA	GAT	GAT + GAT	0.117	0.908	0.818	0.910

Note: Bold indicates the best result in the evaluation metrics. These results are not reported from original studies.