Skip to main content

Table 5 The ablation experimental result on the PDBbind2016 test set

From: DeepTGIN: a novel hybrid multimodal approach using transformers and graph isomorphism networks for protein-ligand binding affinity prediction

Models

R(\(\uparrow\))

RMSE(\(\downarrow\))

MAE(\(\downarrow\))

SD(\(\downarrow\))

CI(\(\uparrow\))

\(\hbox {DeepTGIN}_{without\_pocket}\)

0.720

1.520

1.184

1.506

0.760

\(\hbox {DeepTGIN}_{without\_ligand}\)

0.723

1.512

1.185

1.501

0.763

\(\hbox {DeepTGIN}_{cnn}\)

0.745

1.458

1.132

1.449

0.769

\(\hbox {DeepTGIN}_{lstm}\)

0.725

1.502

1.130

1.495

0.763

DeepTGIN

0.834

1.203

0.949

1.197

0.823

  1. \(\uparrow\) indicates that larger values indicate better performance, while \(\downarrow\) indicates that smaller values indicate better performance
Back to article page

Journal of Cheminformatics

ISSN: 1758-2946

Contact us