Table 5 Performance comparison of different methods on the DS2 Dataset from⁷⁵. We obtained information on 585 drugs for protein sequences, and on 504 drugs for protein structure. Results in The first five rows are taken from⁷⁵.

Substructure-based label propagation model⁴³

0.788

0.208

0.294

0.537

0.197

Vilar’s substructure-based model⁵

0.810

0.244

0.312

0.479

0.232

Classifier ensemble method⁷⁸

0.936

0.487

0.553

0.689

0.462

Weighted average ensemble method⁷⁸

0.646

0.440

0.15

0.226

0.118

NDD⁷⁵

0.994

0.890

0.825

0.804

0.847

BioDKG-DDI⁷⁹

0.967

N/A

0.903

0.918

0.884

DPDDI⁸⁰

0.956

N/A

0.840

0.810

0.754

PS3N (Protein Sequence) (ours)

0.998

0.975

0.978

0.987

0.972

PS3N (Protein Structure) (ours)

0.997

0.975

0.978

0.992

0.964

PS3N (Sequence + Structure) (ours)

0.997

0.970

0.977

0.987

0.970