Research map / Graph learning / Graph self-supervised learning
Hierarchical Generation of Molecular Graphs using Structural Motifs
Wengong Jin, Dr.Regina Barzilay, Tommi Jaakkola
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Generate polymers with large structural motifs rather than atom or small-substructure building blocks | HierVAE · Polymer test set | reconstruction accuracy: 58.5% -> 79.9% | Improved |
| Restrict motif vocabulary to single rings and bonds (ablation) | HierVAE · Polymer test set | reconstruction accuracy: 79.9% -> 71.0% | Hurt |
| Use structural motifs for molecular graph translation instead of atom-based generation | HierG2G · DRD2 | success rate: 75.0% -> 85.9% | Improved |
| Use motif-based decoding instead of an atom-based decoder (ablation) | HierG2G · QED | success rate: 76.1% -> 76.9% | Improved |
| Use motif-based decoding instead of an atom-based decoder (ablation) | HierG2G · DRD2 | success rate: 75.0% -> 85.9% | Improved |
| Remove the top motif layer from the hierarchical encoder (ablation) | HierG2G · QED | success rate: 76.9% -> 75.8% | Hurt |
| Remove the top motif layer from the hierarchical encoder (ablation) | HierG2G · DRD2 | success rate: 85.9% -> 83.5% | Hurt |
| Remove both motif and attachment layers from the encoder (ablation) | HierG2G · QED | success rate: 76.9% -> 67.8% | Hurt |
| Remove both motif and attachment layers from the encoder (ablation) | HierG2G · DRD2 | success rate: 85.9% -> 74.1% | Hurt |
Closest papers by meaning
- DMol: A Highly Efficient and Chemical Motif-Preserving Molecule Generation Platform (NeurIPS 2025 · similarity 0.90)
- Representing Molecules as Random Walks Over Interpretable Grammars (ICML 2024 · similarity 0.88)
- MAGE: Model-Level Graph Neural Networks Explanations via Motif-based Graph Generation (ICLR 2025 · similarity 0.88)
- Graph Generation with $K^2$-trees (ICLR 2024 · similarity 0.88)
- Graph Generative Pre-trained Transformer (ICML 2025 · similarity 0.88)
- DiffMS: Diffusion Generation of Molecules Conditioned on Mass Spectra (ICML 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
