摘要：Practical Molecular Optimization（PMO）基准对分子优化的评估进行了标准化，但其基于简单的性质导向 oracle，无法反映结构导向药物设计。随着该领域转向基于对接的目标函数，评估实践变得缺乏统一标准：不同方法在不同的对接任务上报告结果，且往往没有用于公平比较或样本效率约束的共享协议。为解决这一问题，我们提出面向对接的实用分子优化基准（Practical Molecular Optimization for Docking，PMO-Dock）：这是一套用于基于对接优化的基准与协议，包含 25 项任务，覆盖命中化合物生成、先导化合物优化，以及一项新的特异性任务——该任务要求实现较强的靶点结合，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优，而测试任务被保留用于最终评估，且不在开发过程中使用。测试任务的评估在严格的 oracle 调用预算下进行，以反映真实的药物发现环境。我们对四种具有代表性的高性能方法进行了基准测试，它们涵盖不同优化范式：Saturn（强化学习）、GenMol（离散扩散）、Genetic-guided GFlowNet，以及 ChemLlama（基于 LLM 的进化搜索）。结果表明，没有一种方法能在所有任务中普遍表现最佳；对超参数迁移（跨任务及跨类别）和敏感性的进一步分析显示，迁移通常较为有限，且因方法而异，这推动了依赖具体方法的选择策略。该基准为在受限测试时 oracle 预算下评估模型对未见任务的泛化能力和样本效率，提供了具体的标准。

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.