Post-filtering for residue restraints


Is there a way to do blind (ab initio) docking in HADDOCK and then post-filter (rescore) top models based on residue restraints? Instead of biasing the orientations of the structures for rigid-body docking based on restraints. This would be especially helpful when one is not very sure about the residue restraints.

Dear stianale,

Thank for your interest in using haddock for your research.

First of all, what version of HADDOCK are you using ?

  • If you are using the haddock2.4 webserver, it will be hard to do it.
  • If you are using the beta version of haddock3, this can be performed by adding an emref module and setting all weights parameters to 0 with the exception of w_air to a positive value. This trick will turn on only the ambiguous restraints component in the haddock scoring function and therefore give you a ranking based on satisfied restraints.

You could also consider to run a docking with your current very-ambiguous-restraints, and check if some of the structures resemble the top-ranked ab initio docking.

With the hope this answer will help you

Just to add a piece of detail about this: we’re adding this scoring function to the new version HADDOCK, haddock3, but this addition is still under development. Check the repository for updates!
Another way of doing this would be of course through visual inspection (e.g. using pymol)

Thanks for great answers.

I might try to do both, blind and directed.

I am using two representative states from each of the 10 lowest-frequency non-trivial normal modes of both my protein and DNA, so there will be many docking simulations, even when using ensembles. I guess blind docking would just increase the computational time and cost, because it requires sampling of more rigid bodies. But my reasoning is that ab initio docking and then post-filtering for residues is very powerful.