Radial Distribution Function descriptors for predicting affinity for vitamin D receptor.

The QSAR is an alternative method for the research of new and better Vitamin D analogues with affinity for the VDR receptor. This paper describes the results of applying the Radial Distribution Function (RDF descriptors) approach for predicting the VDR affinity of 38 vitamin D analogues. The model described 80% of the experimental variance, with a standard deviation of 0.35. Leave-one-out, bootstrapping and external set validation were carried out with the aim of evaluating the predictive power of the model. The values of their respective squared correlations coefficients were 0.72, 0.70 and 0.79. The RDF approach was compared with four other predictive models, but none of these could explain more than 71.0% of the variance with six variables in their respective models.