This server can be used to predict PPI-hotspotID residues and for a given protein chain, the method it uses is summarized in this pdf. For your target protein provide either a single chain PDB file or give the 4 letter PDB code and the chain ID that you would like a prediction for. (Note: Currently we are taking only single chain PDB structures).
If you wish, you can view a sample output that was given from a saved prediction we did earlier for 1A5E chain A. Output from a previous prediction on chain A from 1A5E can be found here.
PPI-hotspotID was created by Jon D. Wright, Chen Yao-Chi, Chen Yu-Hsien, Karen Sargsyan and Carmay Lim and is hosted in the Institute of BioMedical Sciences, Academia Sinica, Taipei, Taiwan.
Code that this server uses can be obtained from github as wrigjz/ppihotspotid and wrigjz/conservation_python.
The original manuscript publication in eLife is available from here, from this website you can download supplementary figures and tables S1 and S2.