Results of molecular docking as descriptors to predict human serum albumin binding affinity

Pharmacokinetic properties of a compound are important in drug discovery and development. These properties are most often estimated from the structural properties of a compound with a structural-activity relationship (QSAR) approach. Rapid advances in molecular pharmacology have characterized a number of important proteins that shape the pharmacokinetic profile of a compound. Previous studies have shown that molecular docking, which is capable of analyzing compound-protein interactions, could be applied to make a categorical estimation of a pharmacokinetic property. The present study focused on the binding affinity of human serum albumin (HSA) as an example to show that docking descriptors might also be useful to estimate the exact value of a pharmacokinetic property. A previously reported dataset containing 94 compounds with logK(HSA) values was analyzed. A support vector regression model based on the docking descriptors was able to approximate the observed logK(HSA) in the training and validation dataset with an R(2)=0.79. This accuracy was comparable to known QSAR models based on compound descriptors. In this case study, it was shown that an account of protein flexibility is essential to calculate informative docking descriptors for use in the quantitative estimation of logK(HSA).