用物理先验提升GNN分子可合成性模型在分布外数据的泛化能力。
Physics-Aware Auxiliary Losses Improve Out-of-Distribution Generalization of a GNN Synthesizability Filter
- 在GNN中加入拓扑复杂度与应变能两个物理辅助损失
- 分布外测试下AUC提升0.0066，显著优于纯统计模型
- 适合关注药物分子生成泛化性的研究者
机器学习药物发现流程越来越依赖生成模型提出远离训练数据的分子，而现有可合成性过滤器（SAScore、SCScore、RAscore、DeepSA）为纯统计方法，在分布外（OOD）场景下性能下降。本文探究是否可利用低成本、闭式物理先验作为图神经网络（GNN）的辅助监督来改善泛化能力。在GINE骨干网络上添加两项辅助损失：基于Bertz指数的拓扑复杂度回归，以及基于MMFF94力场能量的应变能软惩罚。在包含65,177个分子的数据集（HIV、Tox21、COCONUT）上，首先复现了强的分布内基线表现，随后在单源分布外分割（训练集为药物样HIV+Tox21，测试集为天然产物COCONUT）上进行四组消融实验（基线 / +复杂度 / +应变 / +两者），重复5次种子实验并使用配对自助法置信区间。所有含物理先验的变体均在分布外显著优于基线（平均分布外AUC 0.9774）：+复杂度提升Δ=+0.0060（95%置信区间[+0.0023, +0.0102]），+应变提升Δ=+0.0032（[+0.0008, +0.0052]），+两者联合提升Δ=+0.0066（[+0.0038, +0.0093]）；所有区间均不包含零，且联合效果最优。各变体在分布内无差异，说明效果仅在分布外显现。研究强调效果微弱，并报告一个警示性方法论发现：单种子实验呈现非单调结果，无法通过多种子验证。
原文摘要 · Abstract (English)
Machine-learning drug-discovery pipelines increasingly rely on generative models that propose molecules far from the data used to train downstream synthesizability filters. Existing filters (SAScore, SCScore, RAscore, DeepSA) are purely statistical and degrade in exactly this out-of-distribution (OOD) regime. We ask whether cheap, closed-form physical priors, used as auxiliary supervision on a graph neural network (GNN), improve OOD generalization. We add two auxiliary losses to a GINE backbone: a topological complexity regression supervised by the Bertz index, and a strain-energy soft penalty supervised by MMFF94 force-field energy. On a 65,177-molecule corpus (HIV, Tox21, COCONUT) labeled by SAScore thresholds we reproduce a strong in-distribution baseline, then evaluate a 4-way ablation (baseline / +complexity / +strain / +both) on a single-source OOD split (train on drug-like HIV+Tox21, test on COCONUT natural products), repeated over 5 seeds with paired bootstrap confidence intervals. All three physics-aware variants give a small but statistically significant OOD improvement over the baseline (mean OOD AUC 0.9774): +complexity Delta = +0.0060 (95% CI [+0.0023, +0.0102]), +strain Delta = +0.0032 ([+0.0008, +0.0052]), +both Delta = +0.0066 ([+0.0038, +0.0093]); every interval excludes zero, and the combination is best. The variants are indistinguishable in-distribution, so the effect is visible only under OOD evaluation. We are explicit that the effects are modest, and we report a cautionary methodological finding: a single-seed version of this experiment produced a qualitatively different (non-monotone) story that did not survive multi-seed evaluation.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。