Quantum similarity QSAR: Study of inhibitors binding to thrombin, trypsin, and factor Xa, including a comparison with CoMFA and CoMSIA methods

Descriptors derived from molecular quantum similarity (MQS) theory were used to construct three-dimensional quantitative structure–activity relationships (3D QSAR) of a set of benzamidine-type compounds that bind to thrombin, trypsin, and factor Xa. Two different techniques were used: quantum similarity matrices and fragment quantum self-similarities. They were applied to a training set made up of 72 molecules, and the optimal models were used to predict the binding affinities for a test set of 16 compounds. Even though the correlation of the training set is not as good as the one obtained with CoMFA and CoMSIA approaches, the predictive capacity of the models is comparable. In contrast to the grid-based methodologies, MQS constitutes a general statistically unbiased method not dependent whatsoever on the parameters of the lattice or on the relative position of the molecules within a grid box.