Rethinking Molecular OOD Generalization via Target-Aware Source Selection

2026-05-13

AI Summary

This study addresses the generalization bottleneck in molecular property prediction under extreme out-of-distribution (OOD) scenarios in AI-driven drug discovery, where existing methods suffer from performance overestimation and negative transfer due to microscopic semantic overlap and indiscriminate domain alignment. To tackle this, the authors introduce SCOPE-BENCH, a novel evaluation benchmark featuring cluster-level OOD splits based on physicochemical descriptor space, and POMA, a framework that enables target-oriented multi-source knowledge transfer through structure-aware source selection—formulated as a retrieval-composition-adaptation pipeline—and dual-scale (topological and pharmacophoric) domain adaptation. Experiments show that state-of-the-art 3D models exhibit a 5.9× average increase in error on SCOPE-BENCH, whereas POMA reduces mean absolute error by 6.2% on average, with improvements reaching up to 11.2%.

Abstract

Robust prediction of molecular properties under extreme out-of-distribution (OOD) scenarios is a pivotal bottleneck in AI-driven drug discovery. Current scaffold-splitting protocols fail to obstruct microscopic semantic overlap, predisposing models to shortcut learning and overestimating their true extrapolation capability; meanwhile, conventional domain adaptation paradigms suffer under extreme structural shifts, as blindly aligning heterogeneous source libraries injects topological noise and triggers negative transfer. To address these two challenges, scaffold-cluster out-of-distribution performance evaluation benchmark (SCOPE-BENCH), a benchmark built on cluster-level partitioning in an explicit physicochemical descriptor space, is proposed alongside policy optimization for multi-source adaptation (POMA), a framework that formulates knowledge transfer as a retrieve-compose-adapt pipeline: labeled source scaffolds structurally close to the unlabeled target are first identified as proxy targets; a reinforcement learning policy then adaptively selects the optimal source subset from an exponentially large candidate pool; and dual-scale domain adaptation is finally performed at macroscopic topological and microscopic pharmacophore scales. Evaluations show that prediction errors of state-of-the-art 3D molecular models surge by up to 8.0x on SCOPE-BENCH with a mean of 5.9x, while POMA achieves up to an 11.2% reduction in mean absolute error with an average relative improvement of 6.2% across diverse backbone architectures. Code is available at https://anonymous.4open.science/r/Molecular-OOD-Code-73F6.

Problem

• out-of-distribution generalization
• molecular property prediction
• scaffold splitting
• domain adaptation
• drug discovery

Innovation

• out-of-distribution generalization
• molecular property prediction
• domain adaptation
• reinforcement learning
• scaffold clustering