网络毒理方法用于抗生素肝毒性预测的研究  

Predicting the Hepatotoxicity of Antibiotics by Using the Network Toxicology Analysis Method

作　　者：唐甜甜 张永红[1] Tang Tian-tian;Zhang Yong-hong(Chongqing Key Laboratory of Drug Metabolism,School of Pharmacy,Chongqing Medical University,Chongqing 400016)

机构地区：[1]重庆医科大学药学院,药物代谢重庆市重点实验室,重庆400016

出　　处：《国外医药（抗生素分册）》2022年第2期65-69,共5页World Notes on Antibiotics

基　　金：国家自然科学基金(22176020,21707012);重庆市自然科学基金(cstc2020jcyj-msxmX0150)。

摘　　要：目的 本研究将网络毒理分析方法应用于处方抗生素肝毒性的预测,以期发现抗生素导致肝损伤(DILI)的可能机理,从而探索与DILI疾病发生发展机理相关的分子信息。方法 应用网络毒理分析方法计算获得抗生素与疾病模块相关的网络拓扑参数,获得网络拓扑参数描述符,结合分子描述符,应用极端梯度提升(Extreme gradient b oosting,XGB)算法建立肝毒性预测模型对各抗生素的肝毒性进行预测,再利用模型中重要贡献的网络拓扑参数描述符加以分析,来阐明抗生素引起肝毒性的可能机制。结果 基于15个描述符建立的肝毒性预测模型拟合和稳健性好,预测抗生素肝毒性结果可靠(ACC=0.80,SE=0.80,SP=0.80,F1=0.80,AUC=0.84),对重要贡献的描述符分析后,借助灰黄霉素(Griseofulvin)在免疫系统中的细胞因子信号传导这一通路中的基因调控来初步解析了抗生素导致肝损伤的潜在机理。结论基于DILI毒理网络收集的拓扑参数描述符成功构建了肝毒性预测模型,准确预测了收集的抗生素的肝毒性,并依据描述符简单分析了灰黄霉素诱导肝损伤的潜在的发生发展机理。Objective In this study,the network toxicology analysis method is applied to predict the hepatotoxicity of antibiotics,in order to discover the possible mechanism of antibiotic-induced liver injury (DILI) and to explore the molecular information related to the occurrence and development of DILI disease.Methods Apply network toxicology analysis method to calculate network topological parameters related to antibiotics and disease modules,and obtain network topological parameter descriptors.Combined with the molecular structure descriptors,the Extreme Gradient Boosting (XGB) algorithm was used to establish the hepatotoxicity prediction model and the hepatotoxicity of each antibiotic was predicted.Then,analysis the significant contributions descriptors in the model elucidates the possible mechanism of antibiotic-induced hepatotoxicity.Results The hepatotoxicity prediction model established based on 15 M&N descriptors had good fit and robustness,and the prediction results of antibiotic hepatotoxicity were reliable (ACC=0.80,SE=0.80,SP=0.80,F1=0.80,AUC=0.84).After analyzing the important descriptors,the potential mechanism of antibiotic-induced liver injury was preliminarily probed through analyzing the gene regulation of griseofulvin in the pathway of cytokine signaling in the immune system.Conclusion Based on the topological parameter descriptors collected by the DILI toxicology network,a hepatotoxicity prediction model was successfully constructed,which accurately predicted the hepatotoxicity of the collected antibiotics,and the potential occurrence and development mechanism of griseofulvin-induced liver injury was analyzed.

关 键 词：网络毒理分析方法 药物导致肝损伤 网络拓扑参数 XGB算法 

分 类 号：R978.1[医药卫生—药品]