I am having issues with preparing the pdb files for multi chain protein docking. I have an antibody structure which I have reformatted so to have both heavy and light chain and I gave this structure a chain identifier B. Then I have its ligand which I have given I chain identifier A. By default, these complexes come from a modeling step and they have each one its own segid.
How should I write the TBL restraint file? From the example on the web we have:
assi (segid A and resid 10 and name CB) (segid B and resid 200 and name CB) 20.0 10.0 0.0
If I keep the segi it is making issues (I guess because of the reason I’ve said before) so that it is not satisfying the cross-links in the final model. So I was wondering if I can use it in the following way:
assi (chain A and resid 10 and name CA) (chain B and resid 200 and name CA) 35.0 35.0 0.0 (min distance 0, max distance 35 angstroms)??
I am trying using these as ambiguous so that HADDOCK can do random exclusion and checking the box for center of mass restraints. How would you do it?
If you want I can send you the pdb files so that you can check them out and tell me how you’d solve the problem.
Did you use the server for this?
If yes can you share the link to a result page?
How many cross-links do you define and how many are now satisfied?
Note that you could also test first the consistency of your crosslinks using our DISVIS server
I think I solved the problem. I’ve set up the pdb files (and the restraints file) so to have chain B for the antibody and then chain A for the antigen but use the antibody as first molecule to dock. In the preparation step, HADDOCK take the first molecule and assigns it segid A, while the second docked molecule is assigned with segid B. So what I end up having is a wrong cross-links restraints file (sites are basically swapped and I get completely out-of-sense orientations). I think none or almost none were satisfied (for obvious reasons). Now I just swapped the order of molecule selection (antigen first and then antibody) and that seemed to do the trick. Do you recommend using DISVIS to have a list of active/passive residues or just running in a scenario when only cross-links are used as restraints? Do you recommend using them as ambiguous restraints (that’s what I think should be more correct to do) or anambiguous? Also, do you enforce some contacts by applying them as center of mass?
Thanks in advance
Glad to hear you solved your issue.
Do you recommend using DISVIS to have a list of active/passive residues or just running in a scenario when only cross-links are used as restraints?
Just to check the consistency of your restraints prior to docking. If they are all consistent (i.e. you can generate solutions that are consistent with all restraints using DISIVIS), you could input those as unambiguous in HADDOCK.
Also, since you are dealing with an antibody, you could define the CDR loops as active residues (turn off passives for the antibody) and select the option to define the entire surface of the antigen as passive. This would be a better strategy than CM restraints.
Ok, I’ll check DISVIS out. I am obtaining best results by using XL-MS restraints as AIRs and with no center of mass selection or using an active passive strategy (defining the CDR loops as active as you already pointed out).
Thanks for the insightful discussion