Journal of Cheminformatics

Table 3 Performance of the models on the validation databases

From: Application of 3D atom pair map in an attention model for enhanced drug virtual screening

	Test		Validation			Summary
	BindingDB		PDBbind	ChEMBL
	seen	unseen		seen	unseen	Mean	Std
RF_ECFP4	0.907	0.760	0.621	0.735	0.714	0.747	0.104
RF_gnn	0.971	0.568	0.591	0.551	0.490	0.634	0.192
RF_graph2vec	0.978	0.728	0.689	0.774	0.736	0.781	0.114
RF_APM	0.988	0.764	0.857	0.841	0.803	0.851	0.085
DBN	0.898	0.649	0.534	0.739	0.646	0.693	0.136
GNN	0.973	0.862	0.601	0.497	0.510	0.689	0.216
drugVQA	0.956	0.887	0.577	0.573	0.514	0.701	0.204
AttentionSite	0.972	0.884	0.650	0.797	0.687	0.798	0.134
APNet	0.946	0.910	0.762	0.801	0.733	0.830	0.093

Boldface text indicates the best-performing model. Italic text indicates the second best-performing model

Back to article page

ISSN: 1758-2946

Contact us

Submission enquiries: journalsubmissions@springernature.com