Molecular Property Prediction under Structural Shift with Tabular Foundation Models

2026-09-29

AI Summary

This study addresses the challenge of structural shift in molecular property prediction by proposing the MolPAIR framework. This method introduces a pioneering dual tabular foundation model (TFM) architecture that integrates CheMeleon molecular representations with in-context learning. By incorporating an explicit molecular error differencing mechanism, MolPAIR refines initial predictions through synergistic comparison between global and differential prediction models, thereby achieving gradient-free fine-tuning without parameter updates. Evaluated across 58 benchmark tasks, the framework yields significant performance improvements on 46 of them, demonstrating the effectiveness of the molecular pair comparison strategy. Ultimately, this work establishes a novel paradigm for enhancing the structural generalization capability of models under frozen-weight settings.

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.

Problem

• Molecular Property Prediction
• Structural Shift
• Tabular Foundation Models
• Structural Generalization

Innovation

• Tabular Foundation Models
• In-context Learning
• Structural Generalization
• Molecular Property Prediction
• MolPAIR