Pang Xudi, He Xuejun, Yang Ying, Wang Ling, Sun Yuzhen, Cao Huiming, Liang Yong
Hubei Key Laboratory of Environmental and Health Effects of Persistent Toxic Substances, School of Environment and Health, Jianghan University, Wuhan 430056, China.
Hubei Key Laboratory of Environmental and Health Effects of Persistent Toxic Substances, School of Environment and Health, Jianghan University, Wuhan 430056, China.
Environ Int. 2025 Jan;195:109244. doi: 10.1016/j.envint.2024.109244. Epub 2024 Dec 28.
Chemically induced neurotoxicity is a critical aspect of chemical safety assessment. Traditional and costly experimental methods call for the development of high-throughput virtual screening. However, the small datasets of neurotoxicity have limited the application of advanced deep learning techniques. The current study developed a hybrid deep learning architecture, NeuTox 2.0, through multimodal feature fusion for enhanced prediction accuracy and generalization ability. We incorporated transfer learning based on self-supervised learning, graph neural networks, and molecular fingerprints/descriptors. Four datasets were used to profile neurotoxicity; these were related to blood-brain barrier permeability, neuronal cytotoxicity, microelectrode array-based neural activity, and mammalian neurotoxicity. Comprehensive performance evaluations demonstrated that NeuTox 2.0 has relatively higher predictive capability across all statistical metrics. Specifically, NeuTox 2.0 exhibits remarkable performance in three of the four datasets. In the BBB dataset, although it does not outperform the PaDEL descriptor model, its performance closely approximates that of the top single-modal model. The ablation experiments indicated that NeuTox 2.0 can learn the deeper structural differences of molecules from various feature extractions and capture complex interactions and mapping relationships between various modalities, thereby improving performance for neurotoxicity prediction. Evaluations of anti-noise ability indicated that NeuTox 2.0 has excellent noise resistance relative to traditional machine learning. We applied the NeuTox 2.0 model to predict the neurotoxicity of 315,790 compounds in the REACH database. The results showed that 701 compounds exhibited potential neurotoxicity in the four neurotoxicity-related predictions. In conclusion, NeuTox 2.0 can be used as an efficient tool for early neurotoxicity screening of environmental chemicals.
化学诱导的神经毒性是化学安全评估的一个关键方面。传统且昂贵的实验方法促使了高通量虚拟筛选技术的发展。然而，神经毒性的小数据集限制了先进深度学习技术的应用。当前的研究通过多模态特征融合开发了一种混合深度学习架构NeuTox 2.0，以提高预测准确性和泛化能力。我们纳入了基于自监督学习的迁移学习、图神经网络以及分子指纹/描述符。使用了四个数据集来描述神经毒性；这些数据集与血脑屏障通透性、神经元细胞毒性、基于微电极阵列的神经活动以及哺乳动物神经毒性有关。全面的性能评估表明，NeuTox 2.0在所有统计指标上都具有相对较高的预测能力。具体而言，NeuTox 2.0在四个数据集中的三个数据集上表现出色。在血脑屏障数据集中，尽管它没有超过PaDEL描述符模型，但它的性能与顶级单模态模型非常接近。消融实验表明，NeuTox 2.0可以从各种特征提取中学习分子更深层次的结构差异，并捕捉各种模态之间复杂的相互作用和映射关系，从而提高神经毒性预测的性能。抗噪声能力评估表明，相对于传统机器学习，NeuTox 2.0具有出色的抗噪声能力。我们应用NeuTox 2.0模型预测了REACH数据库中315,790种化合物的神经毒性。结果表明，在与神经毒性相关的四项预测中，有701种化合物表现出潜在的神经毒性。总之，NeuTox 2.0可以用作环境化学品早期神经毒性筛选的有效工具。
