A nonlinear QSAR Study using Oscillating Search and SVM as An Efficient Algorithm to Model the Inhibition of Reverse Transcriptase by HEPT Derivatives.

Ahmed Allali; Fouad Ferkous; Khaireddine Kraim; Youcef Saihi; Mohammed Brahimi; Faouzi Zaiz; Ouassila Attoui-Yahia

No Detail Found!

A nonlinear QSAR Study using Oscillating Search and SVM as An Efficient Algorithm to Model the Inhibition of Reverse Transcriptase by HEPT Derivatives.

Author(s)

Ahmed Allali Laboratory of Applied Organic Chemistry, Chemistry Department, Badji Mokhtar Annaba University, Algeria. Biology Sciences Department, Hamma Lakhdar El-Oued University, Algeria.

Fouad Ferkous Laboratory of Applied Organic Chemistry, Chemistry Department, Badji Mokhtar Annaba University, Algeria.

Khaireddine Kraim Laboratory of Applied Organic Chemistry, Chemistry Department, Badji Mokhtar Annaba University,Algeria,High School of Technological Education (ENSET) Skikda, Algeria.

Youcef Saihi Laboratory of Applied Organic Chemistry, Chemistry Department, Badji Mokhtar Annaba University, Algeria.

Mohammed Brahimi Department of Computer Science, USTHB University, Algiers, Algeria. Department of Computer Science, Mohamed El Bachir El Ibrahimi University, Bordj Bou Arreridj, Algeria

Faouzi Zaiz Computer Sciences Department; Hamma Lakhdar El-Oued University, Algeria.

Ouassila Attoui-Yahia Laboratory of Applied Organic Chemistry, Chemistry Department, Badji Mokhtar Annaba University, Algeria.

Abstract

Summary: Quantitative structure-activity relationships were constructed for 107 inhibitors of HIV-1 reverse transcriptase that are derivatives of 1-[(2-hydroxyethoxy)methyl]-6-(phenylthio)thymine (HEPT). A combination of a support vector machine (SVM) and oscillating search (OS) algorithms for feature selection was adopted to select the most appropriate descriptors. The application was optimized to obtain an SVM model to predict the biological activity EC50 of the HEPT derivatives with a minimum number of descriptors (SpMax4_Bh (e) MLOGP MATS5m) and high values of R2 and Q2 (0.8662, 0.8769). The statistical results showed good correlation between the activity and three best descriptors were included in the best SVM model. The values of R2 and Q2 confirmed the stability and good predictive ability of the model. The SVM technique was adequate to produce an effective QSAR model and outperformed those in the literature and the predictive stages for the inhibitory activity of reverse transcriptase by HEPT derivatives.

Publication Details

Page(s) 24-32

DOI DOI not available

Published Journal: Journal of Chemical Society of Pakistan, Volume: 40, Issue: 1, Year: 2018

Request the full text from PASTIC

Citations

Downloads

110

Views