机器学习预测内分泌干扰物水生生物毒性效应
Predict Toxicity Effects of Endocrine Disruptor Chemicals on Aquatic Organisms Using Machine Learning
-
摘要: 内分泌干扰物(endocrine disruptor chemicals, EDCs)繁殖毒性实验的周期长、费用高，导致水生生物繁殖毒性数据相对匮乏，限制了EDCs的生态风险评估和管理。毒性数据的预测是解决上述问题的重要手段，也是生态毒理学领域研究的热点和难点之一。在综述国内外利用机器学习预测化学物质的水生生物毒性效应研究的基础上，采用支持向量机(support vector machine, SVM)模型与线性神经网络(linear neural network, LNN)模型，根据定量构效关系(quantitative structure-activity relationship, QSAR)方法对黑头软口鲦(Pimephales promelas)繁殖毒性数据集构建了毒性效应二元分类预测模型，并进行了模型验证与评估。文献分析可知，在使用机器学习预测化合物水生生物毒性效应的研究中，SVM应用最广泛，其次是线性回归与神经网络等；预测急性毒性的研究要多于慢性毒性；分子描述符的筛选没有明确的理论指导，通常为经验与算法相结合，其中与辛醇-水分配系数相关的分子描述符一般具有较高的重要性。实验研究结果表明，经过筛选得到4种描述符作为模型输入变量，描述符分别与原子质量、极化率、电离势和键序有关；SVM对训练集与测试集的预测准确率分别为0.91与0.88，根据受试者工作特征(receiver operating characteristic, ROC)曲线得到的训练集与测试集曲线下面积(area under curve, AUC)分别为0.93与0.88；LNN对训练集与测试集的预测准确率均为0.82，AUC分别为0.87与0.88，表明2个模型均具有良好的泛化与预测能力。SVM的结果优于LNN，表明SVM更适合小样本数据建模。本研究结果可为EDCs的生态毒理学研究及毒性数据的丰富提供重要补充，为EDCs生态风险管控提供科学参考。Abstract: The time-consuming and high costs of reproductive toxicity test of endocrine disruptor chemicals (EDCs) lead to a relatively lack of reproductive toxicity data for aquatic species, which restrict the ecological risk assessment and management of EDCs. The prediction of toxicity data is one of the important methods to solve the above problems, and it is also one of the hotspots and difficulties in the field of ecotoxicology. Based on the review of related research using machine learning to predict chemicals’ toxicity effects on aquatic organisms, a support vector machine (SVM) and a linear neural network (LNN) coupled with quantitative structure-activity relationship (QSAR) were used respectively, to build binary classification models to predict reproduction toxicity for Pimephales promelas, and the models were validated and evaluated using the reproduction toxicity dataset. The results of review showed that SVM was the most widely used model to predict the toxicity effects of compounds on aquatic organisms, followed by linear regression and neural network. Acute toxicity has been studied more than chronic toxicity in application of the machine learning. There was no clear theoretical guidance for the selection of molecular descriptors subset in the field of QSAR. Generally, the combination of experiences and algorithms was applied to filtrate molecular descriptors. The descriptors related to octanol-water partition coefficient were considered to be of high importance. The experimental results are as follows: four descriptors that related to atomic mass, polarizability, ionization potential and bond order were obtained as input variables. The prediction accuracies of SVM for the training set and the test set are 0.91 and 0.88 respectively, and the area under the curve (AUC) of the training set and the test set obtained from the receiver operating characteristic (ROC) curve are 0.93 and 0.88 respectively. The accuracies of LNN for the training set and the test set are both 0.82, and the AUC are 0.87 and 0.88, respectively, indicating that LNN and SVM have good generalization and prediction ability. The results of SVM are better than that of LNN, which means that SVM is more suitable for small dataset. The results can provide an important supplement for the ecotoxicological studies of EDCs and enrich the toxicity data, as well as provide a scientific reference for the ecological risk management of EDCs.
-
Key words:
- endocrine disruptor chemicals /
- Pimephales promelas /
- chronic toxicity /
- machine learning /
- QSAR
-
-
Warner G R, Mourikes V E, Neff A M, et al. Mechanisms of action of agrochemicals acting as endocrine disrupting chemicals[J]. Molecular and Cellular Endocrinology, 2020, 502:110680 JakopinŽ. Assessment of the endocrine-disrupting potential of halogenated parabens:An in silico approach[J]. Chemosphere, 2021, 264:128447 Chung C, Park J, Song J E, et al. Determinants of protective behaviors against endocrine disruptors in young Korean women[J]. Asian Nursing Research, 2020, 14(3):165-172 Vieira W T, de Farias M B, Spaolonzi M P, et al. Removal of endocrine disruptors in waters by adsorption, membrane filtration and biodegradation. A review[J]. Environmental Chemistry Letters, 2020, 18(4):1113-1143 华江环,韩建,郭勇勇,等.孕激素左炔诺孕酮长期低剂量暴露对雄性斑马鱼的生殖毒性[J].生态毒理学报, 2019, 14(2):176-186 Hua J H, Han J, Guo Y Y, et al. Reproductive toxicity in male zebrafish after long-term exposure to low concentrations of progestin levonorgestrel[J]. Asian Journal of Ecotoxicology, 2019, 14(2):176-186(in Chinese)
Vieira W T, de Farias M B, Spaolonzi M P, et al. Endocrine-disrupting compounds:Occurrence, detection methods, effects and promising treatment pathways-A critical review[J]. Journal of Environmental Chemical Engineering, 2021, 9(1):104558 Bottalico L N, Weljie A M. Cross-species physiological interactions of endocrine disrupting chemicals with the circadian clock[J]. General and Comparative Endocrinology, 2021, 301:113650 杨娜,吴航利,王佳,等.农药类内分泌干扰物对动物生殖系统干扰机制的研究进展[J].延安大学学报:自然科学版, 2020, 39(2):87-91 Yang N, Wu H L, Wang J, et al. Research progress on the interference mechanism of pesticide endocrine disrupting chemicals on animal reproductive system[J]. Journal of Yan'an University:Natural Science Edition, 2020, 39(2):87-91(in Chinese)
Schilling J, Nepomuceno A I, Planchart A, et al. Machine learning reveals sex-specific 17β[WT《Times New Roman》]-estradiol-responsive expression patterns in white perch (Morone americana) plasma proteins[J]. Proteomics, 2015, 15(15):2678-2690 蔡德雷,陈江,傅剑云,等.钱塘江水环境内分泌干扰物污染的研究[J].卫生研究, 2011, 40(4):481-484 Cai D L, Chen J, Fu J Y, et al. Study on contamination of endocrine disrupting chemicals in aquatic environment of Qiantang River[J]. Journal of Hygiene Research, 2011, 40(4):481-484(in Chinese)
余方,潘学军,王彬,等.固相萃取-羟基衍生化-气相色谱/质谱联用测定滇池水体中酚类内分泌干扰物[J].环境化学, 2010, 29(4):744-748 Yu F, Pan X J, Wang B, et al. Determination of phenols in surface water of Dianchi Lake by solid extraction-hydroxyl derivatization-GC/MS[J]. Environmental Chemistry, 2010, 29(4):744-748(in Chinese)
张凤仙,胡冠九,郝英群,等.沿海三市饮用水源水内分泌干扰毒性研究[J].生态毒理学报, 2011, 6(3):241-246 Zhang F X, Hu G J, Hao Y Q, et al. Study on endocrine-disrupting toxicity in drinking water sources of three coastal cities[J]. Asian Journal of Ecotoxicology, 2011, 6(3):241-246(in Chinese)
李金荣,郭瑞昕,刘艳华,等.五种典型环境内分泌干扰物赋存及风险评估的研究进展[J].环境化学, 2020, 39(10):2637-2653 Li J R, Guo R X, Liu Y H, et al. Occurrence and risk assessment of five typical environmental endocrine disruptors[J]. Environmental Chemistry, 2020, 39(10):2637-2653(in Chinese)
孟顺龙,宋超,范立民,等.水体中环境内分泌干扰物(EDCs)污染现状及其对鱼类的生殖危害[J].江苏农业学报, 2013, 29(1):202-208 Meng S L, Song C, Fan L M, et al. Pollution of environmental endocrine disrupting chemicals (EDCs) in water and its adverse reproductive effect on fish[J]. Jiangsu Journal of Agricultural Sciences, 2013, 29(1):202-208(in Chinese)
McTavish K, Stech H, Stay F. A modeling framework for exploring the population-level effects of endocrine disruptors[J]. Environmental Toxicology and Chemistry, 1998, 17(1):58-67 黄合田,杨鸿波,孙晓红,等.水产品中内分泌干扰物残留检测方法研究进展[J].水产科学, 2021, 40(2):285-293 Huang H T, Yang H B, Sun X H, et al. Detection methods of environmental endocrine disrupting chemicals in fishery products:Research progress[J]. Fisheries Science, 2021, 40(2):285-293(in Chinese)
Jin X W, Zha J M, Xu Y P, et al. Derivation of aquatic predicted no-effect concentration (PNEC) for 2,4-dichlorophenol:Comparing native species data with non-native species data[J]. Chemosphere, 2011, 84(10):1506-1511 Huang Q S, Bu Q W, Zhong W J, et al. Derivation of aquatic predicted no-effect concentration (PNEC) for ibuprofen and sulfamethoxazole based on various toxicity endpoints and the associated risks[J]. Chemosphere, 2018, 193:223-229 Khan K, Roy K, Benfenati E. Ecotoxicological QSAR modeling of endocrine disruptor chemicals[J]. Journal of Hazardous Materials, 2019, 369:707-718 Tinkov O V, Grigorev V Y, Razdolsky A N, et al. Effect of the structural factors of organic compounds on the acute toxicity toward Daphnia magna[J]. SAR and QSAR in Environmental Research, 2020, 31(8):615-641 Fan J T, Yan Z G, Zheng X, et al. Development of inter species correlation estimation (ICE) models to predict the reproduction toxicity of EDCs to aquatic species[J]. Chemosphere, 2019, 224:833-839 Papa E, Villa F, Gramatica P. Statistically validated QSARs, based on theoretical descriptors, for modeling aquatic toxicity of organic chemicals in Pimephales promelas(fathead minnow)[J]. Journal of Chemical Information and Modeling, 2005, 45(5):1256-1266 Mao W F, Song Y, Sui H X, et al. Analysis of individual and combined estrogenic effects of bisphenol, nonylphenol and diethylstilbestrol in immature rats with mathematical models[J]. Environmental Health and Preventive Medicine, 2019, 24(1):32 Yangali-Quintanilla V, Sadmani A, McConville M, et al. A QSAR model for predicting rejection of emerging contaminants (pharmaceuticals, endocrine disruptors) by nanofiltration membranes[J]. Water Research, 2010, 44(2):373-384 Kovarich S, Papa E, Gramatica P. QSAR classification models for the prediction of endocrine disrupting activity of brominated flame retardants[J]. Journal of Hazardous Materials, 2011, 190(1-3):106-112 张文灏,陈景文,徐童,等.外源化合物在鱼体内生物半减期的QSAR模型[J].生态毒理学报, 2019, 14(3):90-98 Zhang W H, Chen J W, Xu T, et al. QSAR models for predicting biological half-life of xenobiotics in fish[J]. Asian Journal of Ecotoxicology, 2019, 14(3):90-98(in Chinese)
雷太龙.基于机器学习方法的药物毒性的理论预测研究[D].杭州:浙江大学, 2017:1-18 Lei T L. Theoretical prediction of drug toxicity based on machine learning approaches[D]. Hangzhou:Zhejiang University, 2017:1 -18(in Chinese)
Cipullo S, Snapir B, Prpich G, et al. Prediction of bioavailability and toxicity of complex chemical mixtures through machine learning models[J]. Chemosphere, 2019, 215:388-395 Roohi R, Jafari M, Jahantab E, et al. Application of artificial neural network model for the identification the effect of municipal waste compost and biochar on phytoremediation of contaminated soils[J]. Journal of Geochemical Exploration, 2020, 208:106399 薛同来,赵冬晖,韩菲,等. SVR在城市污水BOD预测中的应用[J].新型工业化, 2019, 9(4):94-98 Xue T L, Zhao D H, Han F, et al. Application of support vector regression machine in BOD prediction of urban sewage[J]. The Journal of New Industrialization, 2019, 9(4):94-98(in Chinese)
Borrero L A, Guette L S, Lopez E, et al. Predicting toxicity properties through machine learning[J]. Procedia Computer Science, 2020, 170:1011-1016 Caldwell D J, Mastrocco F, Hutchinson T H, et al. Derivation of an aquatic predicted no-effect concentration for the synthetic hormone, 17 alpha-ethinyl estradiol[J]. Environmental Science&Technology, 2008, 42(19):7046-7054 刘娜,金小伟,王业耀,等.生态毒理数据筛查与评价准则研究[J].生态毒理学报, 2016, 11(3):1-10 Liu N, Jin X W, Wang Y Y, et al. Review of criteria for screening and evaluating ecotoxicity data[J]. Asian Journal of Ecotoxicology, 2016, 11(3):1-10(in Chinese)
Sheffield T Y, Judson R S. Ensemble QSAR modeling to predict multi species fish toxicity lethal concentrations and points of departure[J]. Environmental Science&Technology, 2019, 53(21):12793-12802 Yang L, Wang Y H, Chang J, et al. QSAR modeling the toxicity of pesticides against Americamysis bahia[J]. Chemosphere, 2020, 258:127217 Nolte T M, Peijnenburg W J G M, Hendriks A J, et al. Quantitative structure-activity relationships for green algae growth inhibition by polymer particles[J]. Chemosphere, 2017, 179:49-56 Yap C W. PaDEL-descriptor:An open source software to calculate molecular descriptors and fingerprints[J]. Journal of Computational Chemistry, 2011, 32(7):1466-1474 In Y Y, Lee S K, Kim P J, et al. Prediction of acute toxicity to fathead minnow by local model based QSAR and global QSAR approaches[J]. Bulletin of the Korean Chemical Society, 2012, 33(2):613-619 Marzo M, Lavado G J, Como F, et al. QSAR models for biocides:The example of the prediction of Daphnia magna acute toxicity[J]. SAR and QSAR in Environmental Research, 2020, 31(3):227-243 Hossain K A, Roy K. Chemometric modeling of aquatic toxicity of contaminants of emerging concern (CECs) in Dugesia japonica and its inter species correlation with Daphnia and fish:QSTR and QSTTR approaches[J]. Ecotoxicology and Environmental Safety, 2018, 166:92-101 Lei B L, Li J Z, Liu H X, et al. Accurate prediction of aquatic toxicity of aromatic compounds based on genetic algorithm and least squares support vector machines[J]. QSAR&Combinatorial Science, 2008, 27(7):850-865 Niu B, Jin Y H, Lu W C, et al. Predicting toxic action mechanisms of phenols using AdaBoost Learner[J]. Chemometrics and Intelligent Laboratory Systems, 2009, 96(1):43-48 Ai H X, Wu X W, Zhang L, et al. QSAR modelling study of the bioconcentration factor and toxicity of organic compounds to aquatic organisms using machine learning and ensemble methods[J]. Ecotoxicology and Environmental Safety, 2019, 179:71-78 Sun L, Zhang C, Chen Y J, et al. In silico prediction of chemical aquatic toxicity with chemical category approaches and substructural alerts[J]. Toxicology Research, 2015, 4(2):452-463 Pedregosa F, Varoquaux G, Gramfort A, et al. Scikit-learn:Machine learning in python[J]. Journal of Machine Learning Research, 2011, 12:2825-2830 Guo H N, Wu S B, Tian Y J, et al. Application of machine learning methods for the prediction of organic solid waste treatment and recycling processes:A review[J]. Bioresource Technology, 2021, 319:124114 李云帆.基于机器学习的石化废气污染物排放预测[D].北京:中国石油大学(北京), 2018:7-10 Li Y F. Prediction of petrochemical waste gas emissions based on machine learning[D]. Beijing:China University of Petroleum (Beijing), 2018:7-10(in Chinese) Cheng F X, Shen J, Yu Y, et al. In silico prediction of Tetrahymena pyriformis toxicity for diverse industrial chemicals with substructure pattern recognition and machine learning methods[J]. Chemosphere, 2011, 82(11):1636-1643 Cao Q Q, Liu L, Yang H B, et al. In silico estimation of chemical aquatic toxicity on crustaceans using chemical category methods[J]. Environmental Science Processes&Impacts, 2018, 20(9):1234-1243 OECD. Guidance document on the validation of (quantitative) structure-activity relationship[(Q) SAR] models[R]. Paris:OECD, 2014 Ertürk M D, Saçan M T, Novic M, et al. Quantitative structure-activity relationships (QSARs) using the novel marine algal toxicity data of phenols[J]. Journal of Molecular Graphics&Modelling, 2012, 38:90-100 He L J, Xiao K Y, Zhou C, et al. Insights into pesticide toxicity against aquatic organism:QSTR models on Daphnia magna[J]. Ecotoxicology and Environmental Safety, 2019, 173:285-292 de Morais e Silva L, Lorenzo V P, Lopes W S, et al. Predictive computational tools for assessment of ecotoxicological activity of organic micropollutants in various water sources in Brazil[J]. Molecular Informatics, 2019, 38(8-9):e1800156 王园宁,刘会会,杨先海.构建有机化合物斑马鱼雌激素干扰效应的二元分类模型[J].生态毒理学报, 2019, 14(4):163-169 Wang Y N, Liu H H, Yang X H. Development of binary classification models for predicting estrogenic activity of organic compounds on zebrafish[J]. Asian Journal of Ecotoxicology, 2019, 14(4):163-169(in Chinese)
米晓希,汤爱涛,朱雨晨,等.机器学习技术在材料科学领域中的应用进展[J].材料导报, 2021, 35(15):15115-15124 Mi X X, Tang A T, Zhu Y C, et al. Research progress of machine learning in material science[J]. Materials Reports, 2021, 35(15):15115-15124(in Chinese)
Song I S, Cha J Y, Lee S K. Prediction and analysis of acute fish toxicity of pesticides to the rainbow trout using 2D-QSAR[J]. Analytical Science and Technology, 2011, 24(6):544-555 Grigor'ev V Y, Razdol'skii A N, Zagrebin A O, et al. QSAR classification models of acute toxicity of organic compounds with respect to Daphnia magna[J]. Pharmaceutical Chemistry Journal, 2014, 48(4):242-245 Su Q, Lu W C, Du D S, et al. Prediction of the aquatic toxicity of aromatic compounds to Tetrahymena pyriformis through support vector regression[J]. Oncotarget, 2017, 8(30):49359-49369 Khan K, Khan P M, Lavado G, et al. QSAR modeling of Daphnia magna and fish toxicities of biocides using 2D descriptors[J]. Chemosphere, 2019, 229:8-17 Boone K S, di Toro D M. Target site model:Predicting mode of action and aquatic organism acute toxicity using Abraham parameters and feature-weighted k-nearest neighbors classification[J]. Environmental Toxicology and Chemistry, 2019, 38(2):375-386 Ren S J. Modeling the toxicity of aromatic compounds to Tetrahymena pyriformis:The response surface methodology with nonlinear methods[J]. Journal of Chemical Information and Computer Sciences, 2003, 43(5):1679-1687 Önlü S, Saçan M T. Toxicity of contaminants of emerging concern to Dugesia japonica:QSTR modeling and toxicity relationship with Daphnia magna[J]. Journal of Hazardous Materials, 2018, 351:20-28 Xia B B, Liu K P, Gong Z G, et al. Rapid toxicity prediction of organic chemicals to Chlorella vulgaris using quantitative structure-activity relationships methods[J]. Ecotoxicology and Environmental Safety, 2009, 72(3):787-794 Meng Y B, Lin B L. A feed-forward artificial neural network for prediction of the aquatic ecotoxicity of alcohol ethoxylate[J]. Ecotoxicology and Environmental Safety, 2008, 71(1):172-186 Agatonovic-Kustrin S, Morton D W, Razic S. In silico modelling of pesticide aquatic toxicity[J]. Combinatorial Chemistry&High Throughput Screening, 2014, 17(9):808-818 Polishchuk P G, Muratov E N, Artemenko A G, et al. Application of random forest approach to QSAR prediction of aquatic toxicity[J]. Journal of Chemical Information and Modeling, 2009, 49(11):2481-2488 Habibi-Yangjeh A, Danandeh-Jenagharad M. Application of a genetic algorithm and an artificial neural network for global prediction of the toxicity of phenols to Tetrahymena pyriformis[J]. Chemical Monthly, 2009, 140(11):1279-1288 Louis B, Agrawal V K. QSAR modeling of aquatic toxicity of aromatic aldehydes using artificial neural network (ANN) and multiple linear regression (MLR)[J]. Journal of the Indian Chemical Society, 2011, 88(1):99-107 Sestraş R E, Jäntschi L, Bolboacǎ S D. Poisson parameters of antimicrobial activity:A quantitative structure-activity approach[J]. International Journal of Molecular Sciences, 2012, 13(4):5207-5229 Sangion A, Gramatica P. Hazard of pharmaceuticals for aquatic environment:Prioritization by structural approaches and prediction of ecotoxicity[J]. Environment International, 2016, 95:131-143 -
点击查看大图
计量
- 文章访问数: 2411
- HTML全文浏览数: 2411
- PDF下载数: 135
- 施引文献: 0
下载:
