Models for the prediction of PPARs agonistic activity of indanylacetic acids

Abstract

With the growth of combinatorial chemistry methods in drug discovery, a huge number of compounds are synthesized and screened in parallel for in vitro pharmacological activity, which surprisingly increased the demands of efficient mathematical models to predict desired biological activity. In the present study, an in silico approach using moving average analysis has been applied to a dataset comprising 73 analogues of indanylacetic acid for development of mathematical models for the prediction of each PPAR subtype as well as pan agonistic activity. The value of majority of molecular descriptors (n = 50) for each analogue in the dataset was computed by means of E-Dragon software (version 1.0). Three molecular descriptors, i.e. eccentric adjacency topochemical index-3, information content index-neighbourhood symmetry of 5-order and 2-path kier alpha-modified shape index, yielded the best PPAR subtype specific and sum/pan models by means of moving average analysis. The overall accuracy of the prediction for all individual mathematical models proposed with regard to PPAR α, γ and δ agonistic activity was found to be ≥87, ≥93 and ≥83 %, respectively. Surprisingly, high predictability of the order of ≥86 % was found in the case of sum/pan models. The statistical significance of models/indices was assessed through intercorrelation analysis, sensitivity, specificity and Matthew’s correlation coefficient. High predictability authenticates proposed models for prediction of each PPAR subtype (α/γ/δ) specific as well as pan agonistic activity.