TABLE 3.

Comparison with four GNN methods in two cell-specific datasets under leave-gene-out evaluation setting.

Model	K562			Jurkat
	ROC-AUC	AUPR	Precision@5	ROC-AUC	AUPR	Precision@10
DDGCN	-	-	-	-	-	-
GCATSL	0.523	0.516	0.528	0.508	0.552	0.521
KG4SL	0.508	0.508	0.515	0.501	0.505	0.318
PiLSL	0.627	0.616	0.611	0.629	0.608	0.667
MVGCN-iSL	0.642	0.623	0.632	0.596	0.643	0.598