Quantitative structure–activity relationship study for predicting activity of some medicine compounds by firefly algorithm
Subject Areas :mehdi nekoei 1 , Fatemeh Shams 2
1 - Islamic Azad University of shahrood branch
2 - Department of Chemistry, Shahrood Branch, Islamic Azad University, Shahrood, Iran
Keywords: Inhibitory effect, Firefly Algorithm, Regression stepwise, Quantitative structure–activity relationship, multiple linear regression,
Abstract :
Quantitative structure–activity relationship models were developed for predicting activity of a series of medicine compounds such as pyridine derivatives. The suitable set of the molecular descriptors was calculated and the important descriptors using the variable selections of the stepwise (SW) and the firefly algorithm (FFA) were selected. The predictive quality of the QSAR models was tested for an external set of compounds by multiple linear regression (MLR). Statistical parameters for SW-MLR and FF-MLR were R2 train = 0.835, 0.859, RMSEP = 0.620, 0.786, REP= 10.72, 10.52 respectively. A comparison between the attained results indicated the superiority of the firefly algorithm over the stepwise method in the feature selection. The predicted results of this study can be used to design new inhibitors of melanoma anti-cancer.
_||_