Skip to main content

Table 3 Performance comparison on the prediction of ADMET

From: Topology-enhanced molecular graph representation for anti-breast cancer drug selection

Model

Dataset

Precision

Recall

F1

AUC

AUPR

SVM

MN

0.7843

0.6709

0.6943

0.7957

0.8209

HOB

0.7733

0.7498

0.7607

0.8104

0.6239

hERG

0.8080

0.7589

0.7791

0.8239

0.8494

CYP3A4

0.8397

0.7998

0.8133

0.8518

0.8591

Caco-2

0.8453

0.7807

0.8068

0.8552

0.7525

BiLSTM

MN

0.8226

0.7310

0.7537

0.8195

0.7731

HOB

0.7462

0.7008

0.7165

0.7711

0.7337

hERG

0.8350

0.7914

0.7968

0.8452

0.8196

CYP3A4

0.8838

0.8627

0.8741

0.9129

0.8952

Caco-2

0.8134

0.7954

0.8021

0.8533

0.8258

Graph-CNN

MN

0.8629

0.8293

0.8461

0.8710

0.8623

HOB

0.8110

0.7635

0.7824

0.8369

0.8061

hERG

0.8495

0.8690

0.8556

0.9081

0.8585

CYP3A4

0.8913

0.8827

0.8840

0.9304

0.8731

Caco-2

0.8479

0.8227

0.8306

0.8740

0.8881

ABCD-GGNN

MN

0.9255

0.9613

0.9430

0.9714

0.9862

HOB

0.8637

0.8804

0.8712

0.9130

0.9273

hERG

0.8914

0.8839

0.8842

0.9303

0.9456

CYP3A4

0.9474

0.9163

0.9355

0.9487

0.9322

Caco-2

0.8828

0.8832

0.8829

0.9296

0.9134

  1. We run all models 10 times and report the mean test precision, recall, F1, AUC, and AUPR