Although HADDOCK score can not be compared between different jobs, I was interesting to know whether a scale is available to asses the quality of the results.
For instance, one job of mine yielded -99.2 in HADDOCK score for the best cluster while another job yielded +160, which is obviously much more positive, but since it is entirely different docking run and no scale is given, I really can’t nullify this result.
If you are dealing with different complexes, you simply can’t compare the scores… No absolute scale here since scoring in docking is a very different problem than predicting the binding affinity of a complex as we have demonstrated for example in:
P. Kastritis and A.M.J.J. Bonvin. Are scoring functions in protein-protein docking ready to predict interactomes? Clues from a novel binding affinity benchmark. J. Proteome Research, 9, 2216-2225 (2010).
But if you are comparing different docking runs of the same complex, then of course you can compare the scores. This will however also depend on the amount of information you put into the docking - i.e. if using different sets of restraints you might then better substract the restraint energy from the score in order to compare them.
Thank you for your quick and detailed answer.
On a different subject - I would like to drive a docking simulation using distance restraints.
For that I would need the Expert mode, right?.
If so, I’d be happy if you could grant me this access.
My username here matches the one on your webserver.
Thank you very much!.
***P.S. : Just realized that Im not so clear with the .CNS format needed for the unambiguous distance restraint - Is there any chance an example may be given?.
Upgraded to expert level.
As for the CNS restraint format, examples are provided in our Nature Protocols 2010 server paper - box 4.
S.J. de Vries, M. van Dijk and A.M.J.J. Bonvin The HADDOCK web server for data-driven biomolecular docking. Nature Protocols, 5, 883-897 (2010). Download the final author version here.
Again, many thanks for your quick response and kind help.