GraphCliff: Short-Long Range Gating for Subtle Differences but Critical Changes
Hajung Kim, Jueon Park, Junseok Choe, Sheunheun Baek, Hyeon Hwang, Jaewoo Kang
定量结构-活动关系假定分子结构与生物活性之间的平稳关系。 然而,定义为具有巨大效力差异的结构相似的化合物的对活动悬崖打破了这种连续性。 最近针对活动悬崖的基准显示,具有扩展连接指纹的经典机器学习模型优于图形神经网络。 我们的分析显示,图嵌入物未能在嵌入空间中充分分离结构相似的分子,因此很难区分结构相似但功能不同的分子。 尽管有这种限制,分子图结构本质上是表达和有吸引力的,因为它们保留了分子拓扑。 为了保留分子作为图形的结构表征,我们提出了一种新的模型GraphCliff,它通过闸门机制整合短期和长期信息。 实验结果表明,GraphCliff持续提高非悬崖和悬崖化合物的性能。 此外,层向式节点嵌入分析揭示了相对于强大的基线图模型的过度平滑和增强的判别能力。
Quantitative structure-activity relationship assumes a smooth relationship between molecular structure and biological activity. However, activity cliffs defined as pairs of structurally similar compounds with large potency differences break this continuity. Recent benchmarks targeting activity cliffs have revealed that classical machine learning models with extended connectivity fingerprints outperform graph neural networks. Our analysis shows that graph embeddings fail to adequately separate st...