This is a web server for the rigid-body docking prediction of protein–protein complex structures using a new version (3.0) of the pyDock scoring algorithm. It makes use of a new custom parallel FTDock implementation, with adjusted grid size for optimal FFT calculations. Given the 3D coordinates of two interacting proteins, pyDockWEB returns the best docking orientations as scored mainly by electrostatics and desolvation energy. Thanks to the technical improvements of the version 3.0 of the pyDock framework, this web application exposes many features and modules of the protocol (sampling, scoring, residue restraints). The web server is free for academics, and had an excellent reception by the community, with more than 6,000 jobs served at the time of writing this report, and with many regular users that produce actual scientific research and cited the article 53 times since its publication in 2013. In the last CAPRI meeting (6th), the server scored second among the rest of web servers participating in the community wide experiment.


This web server facilitates the integration of SAXS data and the protein docking protocol pyDock, which was previously shown to double the predictive success rates of docking. This web tool accepts receptor and ligand PDB structures, and CRYSOL SAXS data as input. For advanced use, there exists the option to use previous docking pyDockWEB results as a starting point. The server is available to anyone, and it is becoming more and more popular within the experimental community, with more than 450 jobs served and 15 citations according to Google Scholar.


Atomic modeling of protein–protein interactions requires the selection of near-native structures from a set of docked poses based on their calculable properties. By considering this as an information retrieval problem, in IRaPPA work, methods developed for Internet search ranking and electoral voting were adapted in a pipeline for the integration of biophysical properties. This approach enhances the identification of near-native structures substantially and the method has been implemented in four different protein-protein docking algorithms, including pyDock. For that purpose, pyDockRescoring server was developed in order to rescore pyDockWEB server results.


This is a web service that compiles a wide variety of biophysical functions and energy potentials that are available in the literature and intrinsically scattered through many publications and web servers. CCharPPI collects or re-implements many of these methods in an easy-to-use web tool. The server is able to calculate up to 108 different energetic descriptors for a given protein-protein complex structural model, and has demonstrated to be a useful tool for developing new methods based on machine-learning techniques. CCharPPI is a popular service with more than 2,650 served jobs at the time and cited 17 times according to Google Scholar metrics.

Protein-RNA Benchmark


A few years ago, the group published the first available benchmark fo protein-RNA docking, comprising a total of 106 protein-RNA complexes. In 71 of them, the unbound coordinates were available for at least one of the molecules. This dataset of protein-RNA complexes has been used to test new and current protein-RNA docking methods. This benchmark was updated to its version 1.1 in September 2015 to fix some errors in renumbering, and it has been used by many groups for testing their new methods (number of citations is 23). OPRA server identifies surface residues likely to bind RNA, based on statistical analysis of protein-RNA complex structures.

Protein-Protein Docking and Affinity Benchmark

The PID group co-authored this new version of the protein-protein docking benchmark which, for the first time, included experimentally-determined binding affinity information. This version 5.0 of the docking benchmark contains new fifty-five new complexes, 35 of which have experimentally measured binding affinities. These updated docking and affinity benchmarks now contain a total of 230 and 179 entries, respectively. This work is an essential tool for method-developers and has been cited 41 times since its publication in 2015.

SKEMPI database

The SKEMPI database contains data on the changes in binding affinity and kinetic rate constants upon mutation for protein-protein interactions for which the structure of the complex is available in the PDB. It has been cited 84 times since its publication in 2012. Version 2.0 is currently in development, in collaboration with Iain Moal (EBI) and Justas Dapkunas (Vilnius Univ.).

Logo IBMB_csic-01