Graph attention convolutional neural network model for chemical poisoning of honey bees’ prediction

作     者：Fan Wang Jing-Fang Yang Meng-Yao Wang Chen-Yang Jia Xing-Xing Shi Ge-Fei Hao Guang-Fu Yang 王凡;杨景芳;汪梦瑶;贾辰阳;石星星;郝格非;杨光富

作者机构：Key Laboratory of Pesticide&Chemical BiologyMinistry of EducationCollege of ChemistryCentral China Normal UniversityWuhan 430079China International Joint Research Center for Intelligent Biosensor Technology and HealthCentral China Normal UniversityWuhan 430079China Collaborative Innovation Center of Chemical Science and EngineeringTianjin 300072China 

出 版 物：《Science Bulletin》 (科学通报（英文版）)

年 卷 期：2020年第65卷第14期

页      面：1184-1191,M0004页

核心收录：

学科分类：090504[农学-特种经济动物饲养（含：蚕、蜂等）] 12[管理学] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 081104[工学-模式识别与智能系统] 08[工学] 0905[农学-畜牧学] 09[农学] 0835[工学-软件工程] 0811[工学-控制科学与工程] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：This work was supported in part by the National Key Research and Development Program of China(2017YFD0200506) the National Natural Science Foundation of China(21837001 and 21907036) 

主　　题：Deep learning Graph attention convolutional neural networks Honey bees toxicity Pesticide Molecular design 

摘      要：The impact of pesticides on insect pollinators has caused worldwide concern. Both global bee decline and stopping the use of pesticides may have serious consequences for food security. Automated and accurate prediction of chemical poisoning of honey bees is a challenging task owing to a lack of understanding of chemical toxicity and introspection. Deep learning(DL) shows potential utility for general and highly variable tasks across fields. Here, we developed a new DL model of deep graph attention convolutional neural networks(GACNN) with the combination of undirected graph(UG) and attention convolutional neural networks(ACNN) to accurately classify chemical poisoning of honey bees. We used a training dataset of 720 pesticides and an external validation dataset of 90 pesticides, which is one order of magnitude larger than the previous datasets. We tested its performance in two ways: poisonous versus nonpoisonous and GACNN versus other frequently-used machine learning models. The first case represents the accuracy in identifying bee poisonous chemicals. The second represents performance advantages. The GACNN achieved ~6% higher performance for predicting toxic samples and more stable with ~7%Matthews Correlation Coefficient(MCC) higher compared to all tested models, demonstrating GACNN is capable of accurately classifying chemicals and has considerable potential in practical *** addition, we also summarized and evaluated the mechanisms underlying the response of honey bees to chemical exposure based on the mapping of molecular similarity. Moreover, our cloud platform(http://***) of this model provides low-cost universal access to information, which could vitally enhance environmental risk assessment.