摘要
【目的】对药物的吸收、分布、代谢、排泄、毒性(Absorption,Distribution,Metabolism,Excretion,Toxicity,ADMET)中的代谢、毒性属性进行建模,用于虚拟筛选中的药物性质评价。【方法】提出一种图注意力网络构建药物ADMET预测模型,基于开放数据库和科学文献的药物ADMET数据构造分子图作为分子结构特征,进一步将提出的模型与三种机器学习模型和两种传统的图神经网络模型进行性能比较。【结果】收集整合得到9个ADMET数据集共计149 457条数据。基于图注意力网络的ADMET预测模型在9个数据集中的平均准确率为0.825、平均F1分数为0.672。与机器学习和图神经网络基线模型相比,所提方法在平均准确率和平均F1分数指标上最大提升幅度达6.4%和26.0%。【局限】数据清洗步骤可以精细化处理,模型预测性能可以通过改进预训练策略进一步提升。【结论】所提图注意力网络模型在药物ADMET分类预测上取得良好性能,可将其应用于虚拟药物筛选流程,为计算机辅助药物设计和药物发现提供参考。
[Objective] This study builds a prediction model for drugs’ ADMET properties(Absorption,Distribution, Metabolism, Excretion, Toxicity), aiming to evaluate drugs in virtual screening. [Methods] We constructed a drug ADMET prediction based on the Graph Attention Network(GAN). Then, we used the drug ADMET properties from open access databases and scientific publications to create their molecular graphs and structures. Finally, we compared the GAN-based model with three machine learning models and two graph neural network models. [Results] We collected 9 datasets with 149 457 ADMET records. The proposed prediction model had an average accuracy of 0.825 and an average F1-Score of 0.672 with the 9 datasets, which were 6.4% and26.0% higher than those of the baseline models. [Limitations] The data cleansing process needs to be refined,while the prediction performance can be further improved with a pre-training architecture. [Conclusions] The proposed model could effectively predict a drug’s ADMET, which could help virtual drug screening and computeraided drug developments.
作者
顾耀文
张博文
郑思
杨丰春
李姣
Gu Yaowen;Zhang Bowen;Zheng Si;Yang Fengchun;Li Jiao(Institute of Medical Information,Chinese Academy of Medical Sciences,Beijing 100020,China;XtalPi AI Research Center,Beijing 100089,China)
出处
《数据分析与知识发现》
CSSCI
CSCD
北大核心
2021年第8期76-85,共10页
Data Analysis and Knowledge Discovery
基金
国家自然科学基金(项目编号:81601573)
国家重点研发计划项目(项目编号:2016YFC0901901)的研究成果之一。
关键词
图神经网络
图注意力网络
多源异构数据
ADMET
虚拟筛选
Graph Neural Network
Graph Attention Network
Multi-source Heterogeneous Data
ADMET
Virtual Screening
