Skip to main content

Table 5 The prediction performance of binding affinity

From: DLM-DTI: a dual language model for the prediction of drug-target interaction with hint-based learning

Dataset

Model

AUROC

AUPRC

Sensitivity

Specificity

BIOSNAP

MolTrans

0.895 ± 0.002

0.901 ± 0.004

0.775 ± 0.032

0.851 ± 0.014

Kang et al., S

0.914 ± 0.006

0.900 ± 0.007

0.862 ± 0.025

0.847 ± 0.007

Kang et al., I

0.910 ± 0.012

0.897 ± 0.014

0.830 ± 0.029

0.863 ± 0.011

DLM-DTI, S

0.914 ± 0.003

0.914 ± 0.006

0.848 ± 0.016

0.844 ± 0.024

DLM-DTI, I

0.910 ± 0.005

0.914 ± 0.004

0.850 ± 0.014

0.821 ± 0.006

DAVIS

MolTrans

0.907 ± 0.002

0.404 ± 0.016

0.800 ± 0.022

0.876 ± 0.013

Kang et al., S

0.920 ± 0.002

0.395 ± 0.007

0.824 ± 0.026

0.889 ± 0.015

Kang et al., I

0.942 ± 0.005

0.517 ± 0.017

0.903 ± 0.017

0.866 ± 0.015

DLM-DTI, S

0.895 ± 0.003

0.373 ± 0.017

0.833 ± 0.044

0.802 ± 0.070

DLM-DTI, I

0.898 ± 0.026

0.406 ± 0.026

0.860 ± 0.016

0.786 ± 0.022

BindingDB

MolTrans

0.914 ± 0.001

0.622 ± 0.007

0.797 ± 0.005

0.896 ± 0.007

Kang et al., S

0.922 ± 0.001

0.623 ± 0.010

0.814 ± 0.025

0.916 ± 0.016

Kang et al., I

0.926 ± 0.001

0.639 ± 0.018

0.802 ± 0.022

0.928 ± 0.013

DLM-DTI, S

0.912 ± 0.004

0.643 ± 0.006

0.888 ± 0.014

0.793 ± 0.015

DLM-DTI, I

0.912 ± 0.004

0.636 ± 0.007

0.869 ± 0.023

0.811 ± 0.010

  1. S: single dataset, I: integrated dataset
  2. Performances of five randomly initialized runs were averaged
  3. Best performance is highlighted in bold