Research map / Probabilistic and scientific modeling / Molecules and proteins
Explainable Molecular Property Prediction: Aligning Chemical Concepts With Predictions via Language Models
Zhenzhong Wang, Zehui Lin, Wanyu Lin, Ming Yang, Minggang Zeng, Kay Chen Tan
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 |
|---|---|---|---|
| add the marginal loss L_M (ablation) | Lamole+Bert · seven classification datasets | classification accuracy: 0.7%~3.7% | Improved |
| add the marginal loss L_M (ablation) | Lamole+Bert · Mutag | classification accuracy (%): 84.9 -> 88.2 | Improved |
| add the marginal loss L_M (ablation) | Lamole+Bert · explanation accuracy | explanation accuracy: 1.0%~5.0% | Improved |
| add attention weights to gradient-based explanation (ablation) | Lamole+Bert · explanation accuracy | explanation accuracy: 1.4%~2.3% | Improved |
| add gradients to attention-only explanation (ablation) | Lamole+Bert · explanation accuracy | explanation accuracy: 1.4%~6.3% | Improved |
| increase ground-truth annotation rate from 10% to 20% | Lamole · PTC datasets | explanation accuracy | Improved |
| increase annotation rate from 50% to 100% | Lamole · four PTC datasets | explanation accuracy | No clear effect |
| use only 10% annotated molecules versus none | Lamole · explanation accuracy | explanation accuracy: up to 5% | Improved |
| annotate with ChatGPT instead of human (ablation) | Lamole+Bert · Mutag | explanation accuracy (%): 77.8 -> 67.6 | Hurt |
| annotate with ChemLLM instead of human (ablation) | Lamole+Bert · Mutag | explanation accuracy (%): 77.8 -> 62.5 | Hurt |
| use Group SELFIES input with Lamole instead of SMILES ChemBERTa | Lamole+Bert · PTC-FR | classification accuracy (%): 65.7 -> 70.0 | Improved |
| Lamole+Bert versus GIN on Mutagen | Lamole+Bert · Mutagen | classification accuracy (%): 82.3 -> 74.5 | Hurt |
Closest papers by meaning
- Representing Molecules as Random Walks Over Interpretable Grammars (ICML 2024 · similarity 0.89)
- From Artificially Real to Real: Leveraging Pseudo Data from Large Language Models for Low-Resource Molecule Discovery (AAAI 2024 · similarity 0.88)
- TRIDENT: Tri-Modal Molecular Representation Learning with Taxonomic Annotations and Local Correspondence (NeurIPS 2025 · similarity 0.88)
- Foundation Molecular Grammar: Multi-Modal Foundation Models Induce Interpretable Molecular Graph Languages (ICML 2025 · similarity 0.88)
- Beyond Chemical QA: Evaluating LLM's Chemical Reasoning with Modular Chemical Operations (NeurIPS 2025 · similarity 0.88)
- FGBench: A Dataset and Benchmark for Molecular Property Reasoning at Functional Group-Level in Large Language Models (NeurIPS 2025 · similarity 0.88)
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.
