Conformer datasets for "Equivariant Graph Neural Networks for Toxicity Prediction"

Predictive modeling of toxicity is a crucial step in the drug discovery pipeline. It can help filter out molecules with a high probability of failing in the early stages of de novo drug design. Thus, several machine learning (ML) models have been developed to predict the toxicity of molecules by combining classical ML techniques or deep neural networks with well-known molecular representations such as fingerprints or 2D graphs. But the more natural, accurate representation of molecules is expected to be defined in physical 3D space like in ab initio methods. Recent studies successfully used equivariant graph neural networks (EGNNs) for representation learning based on 3D structures to predict quantum-mechanical properties of molecules. Inspired by this, we investigated the performance of EGNNs to construct reliable ML models for toxicity prediction. We used the equivariant transformer (ET) model in TorchMD-NET for this. Eleven toxicity data sets taken from MoleculeNet, TDCommons, and ToxBenchmark have been considered to evaluate the capability of ET for toxicity prediction. Our results show that ET adequately learns 3D representations of molecules that can successfully correlate with toxicity activity, achieving good accuracies on most data sets comparable to state-of-the-art models. We also test a physicochemical property, namely, the total energy of a molecule, to inform the toxicity prediction with a physical prior. However, our work suggests that these two properties can not be related. We also provide an attention weight analysis for helping to understand the toxicity prediction in 3D space and thus increase the explainability of the ML model. In summary, our findings offer promising insights considering 3D geometry information via EGNNs and provide a straightforward way to integrate molecular conformers into ML-based pipelines for predicting and investigating toxicity prediction in physical space. We expect that in the future, especially for larger, more diverse data sets, EGNNs will be an essential tool in this domain. PAPER https://pubs.acs.org/doi/full/10.1021/acs.chemrestox.3c00032 CODE and MODELS: The conformer data sets and trained toxicity models will be published upon acceptance of this work. The code has been made available at https://github.com/jule-c/ET-Tox, and the processed data as well as pretrained models for training and testing can be downloaded from https://zenodo.org/record/7942946. We can provide the full list of conformers as XYZ files upon request.毒性预测建模是药物研发管线中的关键环节。其可帮助过滤掉在全新药物设计（de novo drug design）早期阶段极有可能失败的分子。为此，已有诸多机器学习（Machine Learning, ML）模型被开发出来，通过将经典机器学习技术或深度神经网络与指纹、二维图等成熟的分子表征相结合，以预测分子毒性。但更为自然且精准的分子表征，理应基于物理三维空间定义，如同从头算（ab initio）方法一般。近期已有研究基于分子三维结构开展表征学习，并成功将等变图神经网络（Equivariant Graph Neural Networks, EGNNs）用于预测分子的量子力学性质。受此启发，我们探究了等变图神经网络构建可靠机器学习毒性预测模型的性能，本次研究采用了TorchMD-NET中的等变Transformer（Equivariant Transformer, ET）模型。我们选用了取自MoleculeNet、TDCommons及ToxBenchmark的11个毒性数据集，以评估等变Transformer用于毒性预测的能力。研究结果表明，等变Transformer能够有效学习分子的三维表征，并可成功关联毒性活性，在多数数据集上均取得了可与当前顶尖模型媲美的优异准确率。我们还测试了一项物理化学性质——分子总能量，以借助物理先验辅助毒性预测。但本研究发现，这两项性质之间并无关联。此外，我们开展了注意力权重分析，以帮助理解三维空间中的毒性预测过程，进而提升机器学习模型的可解释性。综上，本研究通过等变图神经网络融入三维几何信息，所得结论颇具应用前景，并为将分子构象（conformers）集成至基于机器学习的管线中提供了简便途径，以实现物理空间内的毒性预测与研究。我们预期，未来尤其针对规模更大、多样性更强的数据集，等变图神经网络将成为该领域的核心工具。 论文：https://pubs.acs.org/doi/full/10.1021/acs.chemrestox.3c00032 代码与模型： 本研究被录用后，将公开构象数据集与训练完成的毒性预测模型。代码已开源至https://github.com/jule-c/ET-Tox，用于训练与测试的预处理数据及预训练模型可从https://zenodo.org/record/7942946下载。如有需要，我们可提供全部构象的XYZ格式文件。