Newest papers

PMO-Dock: Benchmarking Docking, Specificity, and Generalization in Molecular Optimization

Gor Simonyan, Tatevik Abrahamyan, Narek Abelyan, Tigran Fahradyan, Hrant Khachatrian · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.19011 · cited by 0 works (OpenAlex) · open-access PDF

Spanish

El modelo práctico de optimización molecular (PMO) es una evaluación estandarizada en la optimización molecular, pero se basa en oráculos basados en propiedades simples que no capturan el diseño de fármacos basado en la estructura. A medida que el campo se ha trasladado a objetivos basados en docking, la práctica de evaluación se ha vuelto no estándar, y los diferentes métodos reportan resultados en diferentes tareas de docking, a menudo sin un protocolo compartido para la comparación justa o las restricciones de transferencia de la muestra. Para abordar esto, introducimos el protocolo de tareas Practical Molecular Optimization for Docking (PMO-Dock), un benchmark y protocolo para la optimización basada en docking consistente en 25 tareas que cubren la generación de golpes, la optimización

Machine translation (Meta M2M100); the English original follows.

Abstract

The Practical Molecular Optimization (PMO) benchmark standardized evaluation in molecular optimization, but it is built on simple property-based oracles that do not capture structure-based drug design. As the field has moved to docking-based objectives, evaluation practice has become unstandardized, and different methods report results on different docking tasks, often without a shared protocol for fair comparison or sample-efficiency constraints. To address this, we introduce Practical Molecular Optimization for Docking (PMO-Dock), a benchmark and protocol for docking-based optimization consisting of 25 tasks covering hit generation, lead optimization, and a new specificity task requiring strong on-target binding while penalizing off-target interactions. The protocol separates model development from final evaluation: validation tasks are used for hyperparameter tuning, and test tasks are held out for final evaluation and are not used during development. Evaluation on the test tasks is done under a strict oracle budget to reflect realistic drug-discovery settings. We benchmark four diverse high-performing methods spanning different optimization paradigms, Saturn (reinforcement learning), GenMol (discrete diffusion), Genetic-guided GFlowNet, and ChemLlama (LLM-based evolutionary search). Our results show no universally best method across tasks, additional analyses of hyperparameter transfer (cross-task and cross-category) and sensitivity show that transfer is often limited and varies by method, motivating method-dependent selection strategies. This benchmark provides a concrete standard for evaluating generalization to unseen tasks and sample efficiency under a limited test-time oracle budget.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Computer science, Benchmark (surveying), Oracle, Benchmarking, Machine learning, Generalization

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange