Journal of the Serbian Chemical Society 2013 Volume 78, Issue 4, Pages: 495-506
https://doi.org/10.2298/JSC120713085I
Full text ( 238 KB)
Cited by
QSAR study for anti-HIV-1 activities of HEPT derivatives using MLR and PLS
Ivan Daniela (Romanian Academy, Institute of Chemistry, Department of Computational Chemistry, Timisoara, Romania)
Crisan Luminita (Romanian Academy, Institute of Chemistry, Department of Computational Chemistry, Timisoara, Romania)
Funar-Timofei Simona (Romanian Academy, Institute of Chemistry, Department of Computational Chemistry, Timisoara, Romania)
Mracec Mircea (Romanian Academy, Institute of Chemistry, Department of Computational Chemistry, Timisoara, Romania)
A QSAR study using Multiple Linear Regression (MLR) and a Partial Least
Squares (PLS) methodology was performed for a series of 127 derivatives of
1-(2-hydroxy-ethoxy)methyl]-6-(phenylthio)-timine (HEPT), a potent inhibitor
of the of the human immunodeficiency virus type 1, HIV-1 reverse
transcriptase (RT). To explore the relationship between a pool of HEPT
derivative descriptors (as independent variables) and anti-HIV-1 activity
expressed as log (1/EC50), as dependent variable) MLR and PLS methods have
been employed. Using Dragon descriptors, the present study aims to develop a
predictive and robust QSAR model for predicting anti-HIV activity of the HEPT
derivatives for better understanding the molecular features of these
compounds important for their biological activity. According to the squared
correlation coefficients, which had values between 0.826 and 0.809 for the
MLR and PLS methods, the results demonstrate almost identical qualities and
good predictive ability for both MLR and PLS models. After dividing the
dataset into training and test sets, the model predictability was tested by
several parameters, including the Golbraikh-Tropsha external criteria and the
goodness of fit tested with the Y-randomization test.
Keywords: Golbraikh-Tropsha criteria, dragon descriptors, the Y-randomization
Acknowledgements. This
project was financially supported by Project 1.1 and 1.2 of the Institute of
Chemistry of the Romanian Academy. STATISTICA, MobyDigs and SIMCA-P+
acquisition was funded by Ministerul Educatiei, Cercetarii si Tineretului -
Autoritatea Nationala pentru Cercetare Stiintifica (MedC-ANCS), contract
grant number: 71GR/2006