Research map / Large language models / Agents, code and math
mCLM: A Modular Chemical Language Model that Generates Functional and Makeable Molecules
Carl Edwards, Chi Han, Gawon Lee, Thao Nguyen, Sara Szymkuć, Chetan Kumar Prasad, Bowen Jin, Jiawei Han, Ying Diao, Ge Liu, Hao Peng, Bartosz Andrzej Grzybowski and 2 more
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 GNN encoding of building blocks (ablation) | mCLM · FDA drugs, synthesis-guaranteed blocks | validity (%): 45.21 -> 100.0 | Improved |
| add GNN encoding of building blocks (ablation) | mCLM · FDA drugs, synthesis-guaranteed blocks | makeability (%): 42.8 -> 98.23 | Improved |
| add GNN encoding of building blocks (ablation) | mCLM · FDA drugs, synthesis-guaranteed blocks | synthetic accessibility SA: 2.97 -> 2.43 | Improved |
| add synthesis-guaranteed tokenizer (ablation) | mCLM · FDA drugs, synthesis-guaranteed blocks | synthesizability (%): 73.65 -> 98.23 | Improved |
| add synthesis-guaranteed tokenizer (ablation) | mCLM · FDA drugs, synthesis-guaranteed blocks | synthetic accessibility SA: 3.09 -> 2.43 | Improved |
| add synthesis-guaranteed tokenizer (ablation) | mCLM · FDA drugs, synthesis-guaranteed blocks | validity (%): 100.0 -> 100.0 | No clear effect |
| use building-block tokens with property training instead of atom-level text-based editing | mCLM vs baselines · FDA drugs, synthesis-guaranteed blocks | average improvement (%): 15.0 vs 6.98 (Gemini-2.5-Flash), 2.45 (GPT-4o) | Improved |
| apply mCLM to out-of-distribution drugs with unseen blocks | mCLM · FDA drugs with 3+ blocks (430) | property scores: mean improvement 12.87; PGP 66.0 -> 68.0 | Mixed |
| add mCLM to fallen angel drugs for iterative multi-property refinement | mCLM · Evobrutinib and TNG348 | DILI | Improved |
| use MoleculeSTM for iterative refinement instead of mCLM | MoleculeSTM · Evobrutinib and TNG348 | molecule validity | Hurt |
Closest papers by meaning
- Domain-Agnostic Molecular Generation with Chemical Feedback (ICLR 2024 · similarity 0.89)
- Omni-Mol: Multitask Molecular Model for Any-to-any Modalities (NeurIPS 2025 · similarity 0.88)
- LLM-Augmented Chemical Synthesis and Design Decision Programs (ICML 2025 · similarity 0.88)
- ChemVLM: Exploring the Power of Multimodal Large Language Models in Chemistry Area (AAAI 2025 · similarity 0.88)
- Beyond Chemical QA: Evaluating LLM's Chemical Reasoning with Modular Chemical Operations (NeurIPS 2025 · similarity 0.88)
- Procedural Synthesis of Synthesizable Molecules (ICLR 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.
