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

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.

“实用分子优化”（PMO）基准测试实现了分子优化评估的标准化，但它建立在简单的基于属性的预言机之上，无法捕捉基于结构的药物设计。随着该领域转向基于对接的目标，评估实践变得不再统一，不同的方法在不同的对接任务上报告结果，且往往缺乏公平比较或样本效率约束的共享协议。

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.

为了解决这个问题，我们引入了“对接实用分子优化”（PMO-Dock），这是一个用于基于对接优化的基准测试和协议，包含 25 项任务，涵盖了苗头化合物生成、先导化合物优化，以及一项新的特异性任务，该任务要求强效的靶点结合，同时惩罚脱靶相互作用。该协议将模型开发与最终评估分开：验证任务用于超参数调整，而测试任务则保留用于最终评估，且在开发过程中不予使用。

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).

测试任务的评估是在严格的预言机预算下进行的，以反映真实的药物发现环境。我们对四种跨越不同优化范式的多样化高性能方法进行了基准测试，包括 Saturn（强化学习）、GenMol（离散扩散）、遗传引导的 GFlowNet 以及 ChemLlama（基于大语言模型的进化搜索）。

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.

我们的结果表明，在各项任务中不存在普遍最优的方法。对超参数迁移（跨任务和跨类别）及敏感性的额外分析显示，迁移效果往往有限且因方法而异，这促使人们采取依赖于具体方法的选择策略。该基准测试为评估对未见任务的泛化能力以及在有限测试时间预言机预算下的样本效率提供了一个具体的标准。