Molecular Property Prediction2020en

Open Graph Benchmark - ogbg-molhiv

Molecular property prediction: predict whether a molecule inhibits HIV replication. 41K graphs from MoleculeNet. Binary classification, scaffold split, evaluated by ROC-AUC.

Paper / Website
Current State of the Art

DGN

Research

79.7

roc_auc

OGB ogbg-molhiv — roc_auc

3 results · 1 SOTA advances · higher is better

All results
SOTA frontier
767778798020262027roc_aucDGN

roc_auc Progress Over Time

Showing 2 breakthroughs from May 2022 to Apr 2026

78.779.079.379.579.8May 2022Apr 2026roc_aucDate

Key Milestones

May 2022
GraphGPS

GraphGPS on ogbg-molhiv test ROC-AUC (%). seed — verify

78.8
Source
Apr 2026
DGNCurrent SOTA

Directional Graph Networks on ogbg-molhiv test ROC-AUC (%). seed — verify

79.7
+1.1%
Source
Total Improvement
1.1%
Time Span
4y
Breakthroughs
2
Current SOTA
79.7

Top Models Performance Comparison

Top 3 models ranked by roc_auc

roc_auc1DGN79.7100.0%2GraphGPS78.898.9%3GIN+virtual node77.196.7%0%25%50%75%100%% of best
Best Score
79.7
Top Model
DGN
Models Compared
3
Score Range
2.6

roc_aucPrimary

#ModelScorePaper / CodeDate
1
DGNOpen Source
Research		79.7
editorial
Apr 2026
2
GraphGPSOpen Source
Mila / Valence		78.8
editorial
Apr 2026
3
GIN+virtual nodeOpen Source
Stanford		77.07
editorial
Apr 2026