QSAR Differential Model for Prediction of SIRT1 Modulation using Monte Carlo Method.

Silent information regulator 2 homologue one (SIRT1) modulators have therapeutic potential for a number of diseases like cardiovascular, metabolic, inflammatory and age related disorders. Here, we have studied both activators and inhibitors of SIRT1 and constructed differential quantitative structure activity relationship (QSAR) models using CORAL software by Monte Carlo optimization method and SMILES notation. 3 splits divided into 3 subsets: sub-training, calibration and test sets, were examined and validated with a prediction set. All the described models were statistically significant models. The values of sensitivity, specificity, accuracy and Matthews' correlation coefficient for the validation set of best model were 1.0000, 0.8889, 0.9524 and 0.9058, respectively. In mechanistic interpretation, structural features important for SIRT1 activation and inhibition have been defined.