G protein-coupled receptors (GPCRs) are the most fruitful family of targets for drug discovery. They provided many life-saving medicines in the last decades. Despite the great effort applied to the study of this protein family, structure determination of GPCRs faces high failure rates, since the receptors are very unstable and have intrinsic plasticity. In order to obtain high-resolution crystals, GPCR engineering is usually required to minimise conformational heterogeneity and maximise crystal contacts and stability. Here, we propose a novel approach to the study of the impacts of mutations on GPCR stability, a model capable of predicting effects of mutations on GPCR structure stability without compromising function.

We collected information about mutation on GPCRs and their known influence in protein stability. Subsequently, a supervised machine learning regressor was built, taking use of a mutation modelling approach that uses graph-based signatures and auxiliary features. These combined information is capable to describing protein geometry, physicochemical properties and interatomic interactions and were used for training machine learning supervised model. These model was incorporated in this web-server to provide rapidly and easily tool to evaluate effects of mutations in GPCRs stability.

      Prediction