Prediction of binding energies upon mutation in 3D-structure-known complexes through PyDock scoring functions

The main objective of this work is to build a simple but robust predictor of binding energy changes upon mutation, once structural information is provided. Three different tools are being employed: First, given the wild type structure of a complex, mutations are modelled with Modeller, a powerful program extensively used in protein homology modeling. Using the workframe provided by the tool, 50 different models are created for each mutation. This model diversity helps to take into account protein flexibility and explore, more efficiently, the conformational space of interactions.Then, models are evaluated using Modeller DOPE assessment tool and pyDock scoring function. DOPE [7] is statistical-potential-based tool that evaluates the quality of a model. The pyDock scoring function [8] is formed by different energy terms (electrostatic, desolvation and van der Waals), and was originally designed to deal with protein-protein docking problems. However, due to its energetic basis, we wanted to test their ability to evaluate changes in binding energies