Fork me on GitHub

ProPairs: A Data Set for Protein-Protein Docking

ProPairs provides a set of protein-protein docking complexes with their unbound structures.
The set can be used to benchmark or to test docking algorithms.


For a multi-chain protein complex different partitions into two binding partners are possible. Here, we use the existence of corresponding unbound structures in the PDB as the criterion for legitimate partitions of the considered protein complexes.


Apr/17 2018 Updated data set (PDB snapshot: Jan/2018) with additional 308 novel, non-redundant interfaces (105 with two unbound structures)
Jun/16 2017 Updated data set (PDB snapshot: Jan/2017) with additional 331 novel, non-redundant interfaces (110 with two unbound structures) + updated webpage (Commit)
Apr/26 2016 Updated data set (PDB snapshot: Jan/2016) + support interactive filtering by number of S2 bonds (Commit)
Feb/18 2016 ProPairs supports using time-stamped yearly snapshots of the PDB archive. (Commit)
Jul/06 2015 The ProPairs webpage provides the data sets as single ZIP archives containing all PDB files. See "Raw data" for download links. (Commit)


Please send comments and questions to


Krull, F., Korff, G., Elghobashi-Meinhardt, N., & Knapp, E. W. "ProPairs: A Data Set for Protein-Protein Docking" Journal of Chemical Information and Modeling 55 (2015) 1495–1507