MIT releases CheMeleon, a molecular property prediction model for small data
Understand this story
The MIT team developed a pretrained foundation model, CheMeleon, for molecular property prediction under data-scarce conditions. The team says CheMeleon performed best in 58 tests, outperforming existing methods. CheMeleon is pretrained on about 1,600 molecular descriptors of roughly 1 million molecules, built on top of the open-source machine learning package ChemProp maintained by the MIT Green team, and anyone can download it via ChemProp and fine-tune it for their own use. In a molecular property prediction competition held by OpenADMET, 9 of the top 20 teams used CheMeleon.
AI overview based on source reports · Updated 34 min ago
Coverage timeline
Follow the coverage to explore different perspectives.
- MIT News AIHow to predict the behavior of molecules from scant data
MIT researchers developed CheMeleon, a pre-trained foundation model built on the open-source ChemProp package that learns from descriptors of a million molecules in PubChem and can be fine-tuned to predict molecular properties from limited data. In tests across 58 cases it outperformed other foundation models and classical expert-designed models, and nine of the top 20 entries in an OpenADMET contest used it. The findings were published in the Journal of Chemical Information and Modeling by Zalte, Burns, Abreu, Green, and three BASF researchers.
Activity for this story
Not enough continuous observations to show a trend yet.