Ligand-Based Prediction of Active Conformation by 3D-QSAR Flexibility Descriptors and Their Application in 3+3D-QSAR Models

Abstract

A conceptionally new 3D molecular descriptor type and methodology are deduced by simple statistical thermodynamic reasoning, based on the free energy change encountered during a transformation of a conformational ensemble of the ligand to an active conformation. The performance of the descriptor was first tested on 37 endomorphin analogues with mu-opiate activity. The method resulted in predictive 3D-QSAR models, and the active conformation was also predicted. Generally, the methodology can be combined with the traditional 3D-QSAR techniques in a 3+3D-QSAR manner. This feature was tested on a series of 38 PGF2alphaprostaglandin analogues with antinidatory activity; the extent to which the molecular flexibility explains the variation in the biological activity was estimated and the active conformation was predicted. The novel descriptors in combination with the grid-based SOMFA descriptors resulted in 3+3D-QSAR models with good levels of predictivity leading to the approach of separation of the effect of the molecular interaction field of the active conformation and the effect of the conformational free energy loss.