Đang chuẩn bị nút TẢI XUỐNG, xin hãy chờ
Tải xuống
The solubility of organic compounds in water was related to the environmental behaviors. In this work, the solubility values of 27 organic compounds were calculated by using the different molecular descriptors. The quantitative structure-solubility relationships (QSSRs) were constructed by incorporating the multivariable technique and the genetic algorithm. The important molecular descriptors such as logP, SsCH3_acnt, ABSQ, nelem, nrings, SHBa, Gmax, Gmin, Xvp6, and Xvpc4 were selected to construct the linear models QSSRs with the genetic algorithm. | Journal of Science and Technology 54 (4B) (2016) 162-169 CALCULATION OF ENVIRONMENTAL PROPERTIES FOR ORGANIC COMPOUNDS USING QUANTITATIVE STRUCTURESOLUBILITY RELATIONSHIPS Vo Thanh Cong1, Nguyen Minh Quang1, Pham Nu Ngoc Han2, Nguyen Xuan Truong4, Tran Kim Cuong3, Pham Van Tat2, * 1 2 Industrial University of Ho Chi Minh City, 12 Nguyen Van Bao, Go Vap Tp.HCM Faculty of Science and Technology, Hoa Sen University, 8 Nguyen Van Trang, Dist.1, HCMC 3 4 Department of Chemistry, University of Dalat, 01-Phu Dong Thien Vuong, Dalat City University of Natural Resources and Environment, 236B Le Van Si, Dist. Tan Binh, HCMC * Email: vantat@gmail.com Received: 15 August 2016; Accepted for publication: 10 November 2016 ABSTRACT The solubility of organic compounds in water was related to the environmental behaviors. In this work, the solubility values of 27 organic compounds were calculated by using the different molecular descriptors. The quantitative structure-solubility relationships (QSSRs) were constructed by incorporating the multivariable technique and the genetic algorithm. The important molecular descriptors such as logP, SsCH3_acnt, ABSQ, nelem, nrings, SHBa, Gmax, Gmin, Xvp6, and Xvpc4 were selected to construct the linear models QSSRs with the genetic algorithm. The best four-variable linear model QSSR was obtained from those descriptors. The quality of QSSR linear model indicated in statistical values R2training of 96.60, standard error of estimation, SE of 0.2961, F-statistic of 156.0, P-value of 0.0, R2test of 95.02, and RSS value of 2.823. The architecture of neural network I(4)-HL(4)-O(1) with R2training of 99.03 was constructed by the molecular descriptors in the four-variable linear model. The predicted solubility values of organic substances in test group resulting from these models are in good agreement with those from literature. Keywords: QSSRs, molecular descriptors, multiple regression, neural network. 1. INTRODUCTION The solubility property of .
