|
|
| Type |
Poster Presentation |
| Area |
Industrial Chemistry |
| Room No. |
Exhibition Hall 2+3 |
| Time |
10월 20일 (금요일) 13:00~14:30 |
| Code |
IND.P-65 |
| Subject |
Development of predictive QSPR model of octanol-air partition coefficient for pollutants |
| Authors |
Byeong Woo Son, Byeong Hun Lee, Sung Kwang Lee*
Department of Chemistry, Hannam University, Korea
|
| Abstract |
The octanol–air partition coefficient (KOA) is invaluable for predicting the extent to which a substance partitions from the atmosphere to environmental organic phases such as organic constituents of soil, vegetation, aerosol particles, and even indoor carpet. Also this is useful for assessing chemical transport and persistence of chemicals, which are required by regulatory agency to evaluate new and existing chemicals.
In this study, we developed QSPR models for predicting KOA, which their data were collected from literatures. 717 KOA data were divided into training set and external test set, and molecular descriptors were calculated from 2D chemical structure using the PreADMET program. As it is important to select essential features from a large number of molecular descriptors to develop a robust and predictive QSPR model, we compared two kinds of feature selection method the population based forward selection(PBFS) and the genetic algorithm(GA). Multiple linear regression(MLR) and support vector machine(SVM) were used as learning algorithm to develop QSAR model. The chance correlation and predictability of these model were validated by y-scrambling method and external validation. The reliable range of prediction model can be identified from the kNN based applicability domain(AD). |
| E-mail |
thsquddn789@naver.com |
|
120th General Meeting of the KCS