</> Code with Papers by Syntology

Papers › Pre-training Graph Neural Networks on Molecules by Using Subgraph-Conditioned Graph...

Pre-training Graph Neural Networks on Molecules by Using Subgraph-Conditioned Graph Information Bottleneck

20 Feb 2025archive 2025-07-28

Van Thuy Hoang; O-Joun Lee

This study aims to build a pre-trained Graph Neural Network (GNN) model on molecules without human annotations or prior knowledge. Although various attempts have been proposed to overcome limitations in acquiring labeled molecules, the previous pre-training methods still rely on semantic subgraphs, i.e., functional groups. Only focusing on the functional groups could overlook the graph-level distinctions. The key challenge to build a pre-trained GNN on molecules is how to (1) generate well-distinguished graph-level representations and (2) automatically discover the functional groups without prior knowledge. To solve it, we propose a novel Subgraph-conditioned Graph Information Bottleneck, named S-CGIB, for pre-training GNNs to recognize core subgraphs (graph cores) and significant subgraphs. The main idea is that the graph cores contain compressed and sufficient information that could generate well-distinguished graph-level representations and reconstruct the input graph conditioned on significant subgraphs across molecules under the S-CGIB principle. To discover significant subgraphs without prior knowledge about functional groups, we propose generating a set of functional group candidates, i.e., ego networks, and using an attention-based interaction between the graph core and the candidates. Despite being identified from self-supervised learning, our learned subgraphs match the real-world functional groups. Extensive experiments on molecule datasets across various domains demonstrate the superiority of S-CGIB.

PaperPDFCode

Code

NSLab-CUK/S-CGIB mentioned in paperpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Graph ClassificationGraph Neural NetworkGraph RegressionMolecular Property PredictionSelf-Supervised Learning

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Graph Classification Mutagenicity S-CGIB Accuracy 81.12±0.90 #3 of 5 Archive leaderboard report
Graph Classification NCI1 S-CGIB Accuracy 79.75±0.82 #39 of 69 Archive leaderboard report
Graph Classification NCI109 S-CGIB Accuracy 77.54±1.51 #23 of 38 Archive leaderboard report
Molecular Property Prediction BACE S-CGIB ROC-AUC 86.46±0.81 #2 of 20 Archive leaderboard report
Molecular Property Prediction BBBP S-CGIB ROC-AUC 88.75±0.49 #9 of 29 Archive leaderboard report
Molecular Property Prediction ESOL S-CGIB RMSE 0.816±0.019 #16 of 20 Archive leaderboard report
Molecular Property Prediction FreeSolv S-CGIB RMSE 1.648±0.074 #14 of 22 Archive leaderboard report
Molecular Property Prediction HIV S-CGIB ROC-AUC 78.33±1.34 #4 of 4 Archive leaderboard report
Molecular Property Prediction Lipophilicity S-CGIB RMSE 0.762±0.042 #8 of 13 Archive leaderboard report
Molecular Property Prediction MUV S-CGIB ROC-AUC 77.71±1.19 #3 of 5 Archive leaderboard report
Molecular Property Prediction SIDER S-CGIB ROC-AUC 64.03±1.04 #12 of 19 Archive leaderboard report
Molecular Property Prediction Tox21 S-CGIB ROC-AUC 80.94±0.17 #4 of 20 Archive leaderboard report
Molecular Property Prediction ToxCast S-CGIB ROC-AUC 70.95±0.27 #2 of 8 Archive leaderboard report
Molecular Property Prediction clintox S-CGIB ROC-AUC 78.58±2.01 #13 of 20 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

Graph Neural NetworkSET

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections