Journal of Computer Applications ›› 2024, Vol. 44 ›› Issue (9): 2964-2969.DOI: 10.11772/j.issn.1001-9081.2023091286

• Frontier and comprehensive applications • Previous Articles     Next Articles

Molecular toxicity prediction based on meta graph isomorphism network

Yunchuan HUANG1,2, Yongquan JIANG1,2,3(), Juntao HUANG4, Yan YANG1,2,3   

  1. 1.School of Computing and Artificial Intelligence,Southwest Jiaotong University,Chengdu Sichuan 611756,China
    2.Engineering Research Center of Sustainable Urban Intelligent Transportation,Ministry of Education (Southwest Jiaotong University),Chengdu Sichuan 611756,China
    3.Manufacturing Industry Chains Collaboration and Information Support Technology Key Laboratory of Sichuan Province (Southwest Jiaotong University),Chengdu Sichuan 611756,China
    4.School of Life Science and Engineering,Southwest Jiaotong University,Chengdu Sichuan 611756,China
  • Received:2023-09-19 Revised:2023-12-27 Accepted:2024-01-02 Online:2024-03-15 Published:2024-09-10
  • Contact: Yongquan JIANG
  • About author:HUANG Yunchuan, born in 1998, M. S. candidate. His research interests include artificial intelligence, deep learning, molecular property prediction.
    HUANG Juntao, born in 2002. His research interests include bioengineering, molecular property prediction.
    YANG Yan, born in 1964, Ph. D., professor. Her research interests include artificial intelligence, big data analytics and mining, ensemble learning and multi-view learning, cloud computing and cloud services.
  • Supported by:
    National Natural Science Foundation of China(61976247);Fundamental Research Funds for Central Universities(2682021ZTPY110)

基于元图同构网络的分子毒性预测

黄云川1,2, 江永全1,2,3(), 黄骏涛4, 杨燕1,2,3   

  1. 1.西南交通大学 计算机与人工智能学院, 成都 611756
    2.可持续城市交通智能化教育部工程研究中心(西南交通大学), 成都 611756
    3.四川省制造业产业链协同与信息化支撑技术重点实验室(西南交通大学), 成都 611756
    4.西南交通大学 生命科学与工程学院, 成都 611756
  • 通讯作者: 江永全
  • 作者简介:黄云川(1998—),男,四川遂宁人,硕士研究生,主要研究方向:人工智能、深度学习、分子性质预测
    黄骏涛(2002—),男,江苏南通人,主要研究方向:生物工程、分子性质预测
    杨燕(1964—),女,安徽合肥人,教授,博士,CCF杰出会员,主要研究方向:人工智能、大数据分析与挖掘、集成学习与多视图学习、云计算与云服务。
  • 基金资助:
    国家自然科学基金资助项目(61976247);中央高校基本科研业务费专项资金资助项目(2682021ZTPY110)

Abstract:

To obtain more accurate molecular toxicity prediction results, a molecular toxicity prediction model based on meta Graph Isomorphism Network (GIN) was proposed, namely Meta-MTP. Firstly, graph isomorphism neural network was used to obtain molecular characterization by using atoms as nodes, bonds as edges, and molecules as graph structures. The pre-trained model was used to initialize the GIN to obtain better parameters. A feedforward Transformer incorporating layer-wise attention and local enhancement was introduced. Atom type prediction and bond prediction were used as auxiliary tasks to extract more internal molecular information. The model was trained through a meta learning dual-level optimization strategy. Finally, the model was trained using Tox21 and SIDER datasets. Experimental results on Tox21 and SIDER datasets show that Meta-MTP has good molecular toxicity prediction ability. When the number of samples is 10, compared to FSGNNTR (Few-Shot Graph Neural Network-TRansformer) model in all tasks, the Area Under the ROC Curve (AUC) of Meta-MTP is improved by 1.4% and 5.4% respectively. Compared to three traditional graph neural network models, Graph Isomorphism Network (GIN), Graph Convolutional Network (GCN), and Graph Sample and AGgrEgate (GraphSAGE), the AUC of Meta-MTP improves by 18.3%-23.7% and 7.3%-22.2% respectively.

Key words: deep learning, molecular toxicity prediction, meta learning, graph isomorphism network, Transformer

摘要:

为了获得更准确的分子毒性预测结果,提出基于元图同构网络的分子毒性预测模型Meta-MTP。首先,使用图同构神经网络将原子作为节点、键作为边、分子作为图结构,以获取分子表征;使用预训练模型对图同构网络(GIN)初始化,使它获得更好的参数;引入基于分层注意力和局部增强的前馈Transformer;使用原子类型预测和键预测作为辅助任务提取更多的分子内部信息;通过元学习双层优化策略对模型进行训练;最后使用Tox21和SIDER数据集对模型进行训练。实验结果表明,在Tox21和SIDER数据集上,Meta-MTP具有良好的分子毒性预测能力,当样本数为10时,相较于FSGNNTR(Few-Shot Graph Neural Network-TRansformer)模型,Meta-MTP的曲线下面积(AUC)分别提高了1.4%和5.4%,相较于图同构网络(GIN)、图卷积网络(GCN)和GraphSAGE(Graph Sample and AGgrEgate)3种传统的图神经网络模型,Meta-MTP的AUC提高了18.3%~23.7%和7.3%~22.2%。

关键词: 深度学习, 分子毒性预测, 元学习, 图同构网络, Transformer

CLC Number: