Molecular comparisons improve property predictions under structural shifts
Molecular Property Prediction under Structural Shift with Tabular Foundation Models
Machine Learning
Summary
Predicting properties of molecules that look different from what a computer has seen before is hard but important for designing new drugs and materials. The authors study how special computer models called tabular foundation models can learn from examples without changing their settings. They introduce a method called MolPAIR that compares molecules in pairs to refine predictions, improving accuracy on many tasks. Their approach works well across different molecular representations and models, and it keeps the main model unchanged while making better predictions.
What this means in practice
- •For drug discovery teams: Predict properties of new drug-like molecules that differ structurally from training data using improved in-context learning methods.
- •For materials science researchers: Enhance prediction accuracy for novel materials with structural variety by refining molecular property predictions from existing labeled examples.
Authors
Jinmo Lee, Dooho Lee, Minho Jeong, Jaemin Yoo
Abstract
Predicting molecular properties for compounds that differ structurally from labeled training molecules is important for drug discovery and materials design. Tabular foundation models (TFMs) offer a promising approach through in-context learning, but their performance under structural shifts and the value of molecular comparisons in this setting remain underexplored. We study structural generalization in molecular property prediction and introduce MolPAIR (Molecular Pair-Augmented In-context Refinement), a framework that combines molecule-level and molecular-pair contexts without task-specific parameter updates. A global tabular foundation model (TFM) first predicts a query's property from labeled molecular examples. A second frozen TFM predicts differences in prediction errors between the query and labeled reference molecules, using these comparisons to refine the initial prediction. Across 58 MoleculeACE and Polaris tasks, CheMeleon representations combined with TabPFN-3 already outperform each evaluated baseline on a majority of tasks. MOLPAIR further improves this predictor on 46 of 58 tasks, with gains across four molecular representations and three TFM backbones. These results show that explicit molecular comparisons can strengthen tabular in-context learning for structural generalization while keeping the molecular encoder and pretrained model weights fixed. The code and datasets are available at https://github.com/nums-ai/MolPAIR.