Molecular Property Prediction on MoleculeNet MUV (scaffold)
0.903ROC-AUCSYN-FUSION
Evaluation Results
| Method | Links | |
|---|---|---|
| SYN-FUSIONfusion_source=MolCLR2023.08 | 0.903 | |
| SYN-FUSIONfusion_source=Hu et. al2023.08 | 0.889 | |
| DELTABackbone=GIN (supervised_contextpred), Strategy=Fine-tuning strategy2022.03 | 0.859 | |
| Feature (DELTA w/o ATT)Backbone=GIN (supervised_contextpred), Strategy=Fine-tuning strategy2022.03 | 0.856 | |
| L2_SPBackbone=GIN (supervised_contextpred), Strategy=Fine-tuning strategy2022.03 | 0.85 | |
| GTOT-TuningBackbone=GIN (supervised_contextpred), Strategy=GTOT-Tuning2022.03 | 0.85 | |
| MolCLRbackbone=GIN2023.08 | 0.838 | |
| StochNormBackbone=GIN (supervised_contextpred), Strategy=Fine-tuning strategy2022.03 | 0.825 | |
| Hu et al.2023.08 | 0.814 | |
| GINGraph-level pre-training=Supervised, Node-level pre-training=Infomax2019.05 | 0.813 | |
| GINGraph-level pre-training=Supervised, Node-level pre-training=Context Prediction2019.05 | 0.813 | |
| Fine-TuningBackbone=GIN (supervised_contextpred), Strategy=Baseline2022.03 | 0.813 | |
| GINGraph-level pre-training=Supervised, Node-level pre-training=Attribute Masking2019.05 | 0.812 | |
| GINGraph-level pre-training=Supervised, Node-level pre-training=None2019.05 | 0.794 | |
| SimSGT2026.02 | 0.79 | |
| BSSBackbone=GIN (supervised_contextpred), Strategy=Fine-tuning strategy2022.03 | 0.786 | |
| Mole-BERT2023.11 | 0.786 | |
| Mole-BERTPre-training status=Pre-trained, Pre-training dataset=ZINC15, Evaluation protocol=Fine-tuning, Split type=Scaffold-split2024.05 | 0.786 | |
| AttentiveFP2026.06 | 0.786 | |
| GINGraph-level pre-training=Supervised, Node-level pre-training=EdgePred2019.05 | 0.785 | |
| GraphMAE22026.02 | 0.785 | |
| GraSPNet2026.02 | 0.785 | |
| GraphMAE2Pre-training status=Pre-trained, Pre-training dataset=ZINC15, Evaluation protocol=Fine-tuning, Split type=Scaffold-split2024.05 | 0.785 | |
| GraphMVP2023.11 | 0.777 | |
| GraphMVPPre-training status=Pre-trained, Pre-training dataset=ZINC15, Evaluation protocol=Fine-tuning, Split type=Scaffold-split2024.05 | 0.777 | |
| MGSSL2023.11 | 0.776 | |
| Hi-GMAE-FPre-training status=Pre-trained, Pre-training dataset=ZINC15, Evaluation protocol=Fine-tuning, Split type=Scaffold-split2024.05 | 0.775 | |
| MAM2023.11 | 0.774 | |
| TMCL2023.11 | 0.772 | |
| Mole-BERT2026.02 | 0.772 | |
| Hi-GMAE-GPre-training status=Pre-trained, Pre-training dataset=ZINC15, Evaluation protocol=Fine-tuning, Split type=Scaffold-split2024.05 | 0.771 | |
| UnifiedMolPre-training status=Pre-trained, Pre-training dataset=ZINC15, Evaluation protocol=Fine-tuning, Split type=Scaffold-split2024.05 | 0.769 | |
| BGRL2026.02 | 0.767 | |
| JOAO2023.11 | 0.766 | |
| GPSEaugmented2026.06 | 0.766 | |
| MAMEncoder=vanilla VQ-VAE2023.11 | 0.765 | |
| GraphCL2023.11 | 0.764 | |
| GraphMAE2023.11 | 0.764 | |
| AD-GCL2023.11 | 0.763 | |
| GraphMAE2026.02 | 0.763 | |
| MGSSL2026.02 | 0.763 | |
| GraphMAEPre-training status=Pre-trained, Pre-training dataset=ZINC15, Evaluation protocol=Fine-tuning, Split type=Scaffold-split2024.05 | 0.763 | |
| GraphMAE2026.06 | 0.763 | |
| D-MPNN2023.08 | 0.762 | |
| 3D InfoMax2023.11 | 0.762 | |
| GraphLogPre-training status=Pre-trained, Pre-training dataset=ZINC15, Evaluation protocol=Fine-tuning, Split type=Scaffold-split2024.05 | 0.76 | |
| GraphLoG2026.06 | 0.76 | |
| GPT-GNN2023.11 | 0.759 | |
| APTBase Model=Mole-BERT2023.11 | 0.759 | |
| GINGraph-level pre-training=None, Node-level pre-training=Context Prediction2019.05 | 0.758 | |
| AttrMask2023.11 | 0.758 | |
| ContextPredPre-training status=Pre-trained, Pre-training dataset=ZINC15, Evaluation protocol=Fine-tuning, Split type=Scaffold-split2024.05 | 0.758 | |
| SSL pre-trainingPre-training strategy=SSL2026.06 | 0.758 | |
| TMCLMode=w/o Ltri2023.11 | 0.757 | |
| GraphLoG2023.11 | 0.755 | |
| SimGRACE2023.11 | 0.754 | |
| SimGRACEPre-training status=Pre-trained, Pre-training dataset=ZINC15, Evaluation protocol=Fine-tuning, Split type=Scaffold-split2024.05 | 0.754 | |
| GINGraph-level pre-training=None, Node-level pre-training=Infomax2019.05 | 0.753 | |
| InfomaxPre-training status=Pre-trained, Pre-training dataset=ZINC15, Evaluation protocol=Fine-tuning, Split type=Scaffold-split2024.05 | 0.753 | |
| APTBase Model=EdgePred2023.11 | 0.752 | |
| EdgePred2023.11 | 0.751 | |
| Infomax2026.02 | 0.748 | |
| GINGraph-level pre-training=None, Node-level pre-training=Attribute Masking2019.05 | 0.747 | |
| Attr mask2026.02 | 0.747 | |
| AttrMaskingPre-training status=Pre-trained, Pre-training dataset=ZINC15, Evaluation protocol=Fine-tuning, Split type=Scaffold-split2024.05 | 0.747 | |
| G2PT2026.06 | 0.745 | |
| InfoGraph2023.11 | 0.744 | |
| JOAO2026.02 | 0.742 | |
| GINGraph-level pre-training=None, Node-level pre-training=EdgePred2019.05 | 0.741 | |
| GraphCL2026.02 | 0.738 | |
| No Pre-train2026.02 | 0.734 | |
| GCNPre-training status=None, Pre-training dataset=N/A, Evaluation protocol=Supervised Learning, Split type=Scaffold-split2024.05 | 0.732 | |
| RGCLPre-training status=Pre-trained, Pre-training dataset=ZINC15, Evaluation protocol=Fine-tuning, Split type=Scaffold-split2024.05 | 0.731 | |
| G-Motif2023.11 | 0.73 | |
| GraphLOG2026.02 | 0.728 | |
| ContextPred2023.11 | 0.725 | |
| AD-GCLPre-training status=Pre-trained, Pre-training dataset=ZINC15, Evaluation protocol=Fine-tuning, Split type=Scaffold-split2024.05 | 0.722 | |
| G-Contextual2023.11 | 0.721 | |
| GIN2023.08 | 0.718 | |
| GINGraph-level pre-training=None, Node-level pre-training=None2019.05 | 0.718 | |
| MPTr+CIPEBackbone=MPTr, Positional encoding=CIPE2026.06 | 0.718 | |
| JOAOPre-training status=Pre-trained, Pre-training dataset=ZINC15, Evaluation protocol=Fine-tuning, Split type=Scaffold-split2024.05 | 0.717 | |
| GCN2023.08 | 0.716 | |
| TMCLMode=w/o Lcon2023.11 | 0.716 | |
| ADGCL2026.02 | 0.715 | |
| GraphGPSPre-training status=None, Pre-training dataset=N/A, Evaluation protocol=Supervised Learning, Split type=Scaffold-split2024.05 | 0.712 | |
| Hi-GMAE-GPre-training status=None, Pre-training dataset=N/A, Evaluation protocol=Supervised Learning, Split type=Scaffold-split2024.05 | 0.712 | |
| MPTr+HKSEBackbone=MPTr, Positional encoding=HKSE2026.06 | 0.709 | |
| No pretrain2023.11 | 0.707 | |
| SAGPoolPre-training status=None, Pre-training dataset=N/A, Evaluation protocol=Supervised Learning, Split type=Scaffold-split2024.05 | 0.704 | |
| CP2026.02 | 0.702 | |
| MPTrBackbone=MPTr2026.06 | 0.701 | |
| MPTr+RWSEBackbone=MPTr, Positional encoding=RWSE2026.06 | 0.699 | |
| GraphCLPre-training status=Pre-trained, Pre-training dataset=ZINC15, Evaluation protocol=Fine-tuning, Split type=Scaffold-split2024.05 | 0.698 | |
| GraphCLPre-training strategy=GraphCL2026.06 | 0.698 | |
| MAE2026.02 | 0.697 | |
| MPTr+CycleSEBackbone=MPTr, Positional encoding=CycleSE2026.06 | 0.696 | |
| MPTr+LapPEBackbone=MPTr, Positional encoding=LapPE2026.06 | 0.695 | |
| SVM2023.08 | 0.673 | |
| GMTPre-training status=None, Pre-training dataset=N/A, Evaluation protocol=Supervised Learning, Split type=Scaffold-split2024.05 | 0.672 |