分子相似性在药物设计中常失效，该研究揭示并修复了隐藏的预测错误。
When Molecular Similarity Works: Property Cliffs Reveal Hidden Errors
- 提出悬崖分割评估法，暴露分子结构相似但性质突变区域的错误
- 实验显示悬崖区域错误率高出15%，模型误差差距缩小30%
- 适合关注模型可靠性与药物研发安全性的研究者
分子性质准确预测是药物发现和材料设计的基础，但即使最先进的模型仍存在局部失败模式，传统评估指标难以察觉。分子相似性最应发挥作用的区域，恰恰是标准评估最容易误导的地方。性质悬崖现象揭示了这一差距：结构相近的分子可能在目标性质上差异显著，导致整体表现良好却在高风险局部区域失效。为此，本文提出CliffSplit评估协议，构建局部支持且暴露悬崖的测试样本；以及CliffLoss训练阶段的无模型依赖纠错机制。在三个QM9任务和三个MoleculeNet任务上，使用五种骨干网络进行实验，结果表明CliffSplit在悬崖密集的QM9区域揭示至少15%更高的错误率，CliffLoss在Lipophilicity任务上将悬崖区与平稳区误差差距减少30%，并使整体平均绝对误差降低9.7%。这些成果将分子相似性失败从描述性异常转变为可量化的评估基准。代码已公开于https://anonymous.4open.science/r/Cliff_Loss。
原文摘要 · Abstract (English)
Accurate prediction of molecular properties underpins drug discovery and material design, yet even state-of-the-art models remain vulnerable to localized failure modes that aggregate metrics cannot detect. The places where molecular similarity should be most helpful are also places where standard evaluation can be most misleading. Property cliffs expose this gap: structurally similar molecules can still differ sharply in target property, so models with competitive overall performance may fail in high-risk local neighborhoods. To expose and mitigate this failure mode, CliffSplit, a cliff-aware evaluation protocol that constructs locally supported, cliff-exposed test cases, and CliffLoss, a model-agnostic train-only mitigation mechanism for cliff-sensitive errors, are introduced. Experiments on three QM9 targets and three MoleculeNet tasks across five backbones show that CliffSplit reveals at least 15% higher error in cliff-heavy QM9 regions, while CliffLoss reduces the cliff-to-smooth error gap by up to 30% on Lipophilicity and improves overall MAE by 9.7%. Together, these results turn molecular similarity failure from a descriptive anomaly into a benchmarked evaluation problem for molecular machine learning. The code is available at https://anonymous.4open.science/r/Cliff_Loss.
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。