Abstract
In the present study, predictive quantitative structure – activity relationship (QSAR) models for anti-malarial activity of 4-aminoquinolines have been developed. CORAL, which is freely available on internet (http://www.insilico.eu/coral), has been used as a tool of QSAR analysis to establish statistically robust QSAR model of anti-malarial activity of 4-aminoquinolines. Six random splits into the visible sub-system of the training and invisible subsystem of validation were examined. Statistical qualities for these splits vary, but in all these cases, statistical quality of prediction for anti-malarial activity was quite good. The optimal SMILES-based descriptor was used to derive the single descriptor based QSAR model for a data set of 112 aminoquinolones. All the splits had r2> 0.85 and r2> 0.78 for subtraining and validation sets, respectively. The three parametric multilinear regression (MLR) QSAR model has Q2 = 0.83, R2 = 0.84 and F = 190.39. The anti-malarial activity has strong correlation with presence/absence of nitrogen and oxygen at a topological distance of six.
Keywords: Anti-malarial activity, 4-aminoquinolines, QSAR, optimal descriptor, CORAL software.
32