Table 5.

Performance comparison of ablation experiments.

Dataset	Variants	MSE ↓	CI ↑	${\mathit{r}}_{\mathit{m}}^2$↑	Pearson ↑
Davis	Without Motif-Level	0.200	0.903	0.715	0.867
	Without Skip-Connection	0.201	0.903	0.728	0.866
	Without Attension	0.203	0.897	0.709	0.865
	MSGNN-DTA	0.195	0.906	0.719	0.871
KIBA	Without Motif-Level	0.123	0.905	0.790	0.906
	Without Skip-Connection	0.122	0.904	0.788	0.906
	Without Attension	0.124	0.906	0.808	0.905
	MSGNN-DTA	0.117	0.908	0.818	0.910

Note: Bold indicates the best result of the evaluation metrics.