Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
PMO-Dock establishes a standardized benchmark and protocol for evaluating molecular optimization methods using docking-based objectives, addressing the current unstandardized landscape. It introduces 25 tasks including hit generation, lead optimization, and a novel specificity task, evaluated under strict oracle budgets and with held-out test sets to ensure fair comparison and realistic drug-discovery simulation. Benchmarking four diverse methods reveals no single best performer and highlights limited hyperparameter transfer, suggesting the need for method-dependent optimization strategies.
Molecular optimization methods struggle to generalize across docking tasks, with no single approach dominating and hyperparameter transfer proving surprisingly limited.
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.