A multi-step approach including models’ construction (multiple sequence alignment, homology modeling), complex assembling (protein complex refinement with HADDOCK and complex equilibration), and protein-protein interface (PPI) characterization (including both structural and dynamics analysis) were performed. Our database can be easily applied to several GPCR sub-families, to determine the key structural and dynamical determinants involved in GPCR coupling selectivity.

SPOTONE is a new Machine-Learning (ML) predictor able to accurately classify protein Hot-Spots (HS) via sequence-only features. This algorithm shows an accuracy, AUROC, precision, recall and F1-score of 0.82, 0.83, 0.91, 0.82 and 0.85, respectively, in an independent testing set.