【來自SCI的前沿論文】———QSARstudy of halogenated benzene bioaccumulation factors in fish
作（著）者：陳福迪
導師：王斌
學術期刊或出版社名稱：BulgarianChemical Communications
期刊封面：
期刊簡介：BulgarianChemical Communications期刊為SCI-E收錄，收錄化學方向文章。
DOI：WOS:000401034500031
論文主要内容：
Certainelements or compounds, that are difficult to decompose, accumulate inorganisms. The phenomenon of higher compound concentrations inorganisms than in the environment is called bio-concentration. Thisresearch studies bio-concentration factors (BCF) of 21kinds of halogenatedbenzene in fish. Here, 14 kindsof halogenatedbenzene molecule parameters are randomly selected as training setsand the remaining parameters are considered as testing setsto calculate halogenated benzene molecule descriptors. A predictionmodel of the quantitative structure–activity relationship betweenbio-concentration and molecular descriptors is built using themultiple linear regression method. The independent variable withserious collinearity is eliminated to obtain an optimal predictionmodel (R2=0.919).
Fig.1.Comparison of the experimental and predicted values
Theprediction data of linear regression are also obtained using themolecular parameters of the test set. A support vector machine (SVM)is used to predict the result of the test set using the training setas study samples.
Fig.2.Support vectors determining the position of the optimal hyperplane.
Thebest method for predicting the bio-concentration factor ofhalogenobenzene in fish is determined by comparing the predictionaccuracy of the two methods. The final results indicate that themodel officially built by stepwise regression can effectively predictBCF. SVM is more accurate in predicting BCF. SVM possesses betterpredictive capability in issues with a small amount of samples.
作者簡介：陳福迪，伟德 官网海洋生物學2014級研究生，主要研究方向為生态毒理學建模，碩士期間曾獲得2015和2016年研究生國家獎學金，參與項目《基于IPv6的水産品物聯網試驗系統的設計與應用》，畢業論文題目為《有機磷酸酯對雙齒圍沙蠶幼體急性毒性預測模型的研究》。
作者生活照
