Evaluation of molecular structure based descriptors for the prediction of pEC50(M) for the selective adenosine A2A Receptor

Abstract

In this study, QSAR models were built for the prediction of pEC50(M) for the A2A adenosine receptor with the help of a cured experimental dataset of 112 ligands. The molecular structure based Rotatable bonds count descriptor (RotBtFrac), CrippenlogPa&MR descriptor (CrippenLogP) and nhigh lowest atom weighted BCUTS descriptor (BCUTw-1l) showed better response towards the pEC50(M). The one descriptor, two and three descriptors models using the above descriptors displayed satisfactory performance towards the prediction of pEC50(M) for the A2A receptor. The performance of the proposed QSAR models has also been evaluated by the number of statistical parameters, internal as well as external validation techniques. The stability of the model is ascertained by taking three random splits. The QSAR models with three descriptors were found robust, having their (R2 (validation set) between 0.90 to 0.95, R2m(Avr) between 0.72 to 0.93 and ΔR2m between 0.004 to 0.13). The response of the conformers of a given ligand in the binding site of A2A target protein is also studied by molecular docking. The presented models can be used to screen the big databases to generate better leads for A2A adenosine receptor and thus can contribute to deal with different diseases related to A2A receptor.