Protein Protein docking without restraint definitions

HADDOCK
1

Hi Prof. Bonvin,

I have experienced with bound ligand protein-protein docking with restraint definition with 5A active site residue from the ligand. HADDOCK generated almost similar poses as reference. Now, assume if I don’t have information about ligand in the second protein. I don’t know the second protein where exactly bind with the first protein. Blindly, I would like to submit a job. Is it possible to submit a job without defining active and passive residues in HADDOCK interface? If yes then I would like to check do I get similar poses like a reference protein. Looking forward to hear from you.

Regards,

Haresh Ajani

Distorted output structures
2

Well, that’s not the scenario for which HADDOCK was designed.

But there is indeed a way to do it. Two scenarios:

  1. You do know the binding site on protein 1 but no info for protein 2: In that case define active residues for protein 1 and define all solvent accessible residues of protein 2 as passive

  2. You don’t know anything about both proteins: Use in that case either the random AIRs (expert or guru access required) or the centre-of-mass restraints options. Those can be selected in the distance restraints menu:

Random patches
Define randomly ambiguous interaction restraints from accessible residues

or

Center of mass restraints
Define center of mass restraints to enforce contact between the molecules

In both cases do increase the sampling of models for it0 to 10000 and it1/water to 400.

I would also suggest to perform then a more statistical analysis of the contacted residues to try to identify possible binding site. Something similar in described in our protein-ligand tutorial at:

3

Thanks Prof. Bonvin for answer.

4

Hi. With reference to this. I have a question. Say one antibody (A) binds to domain 2 of a protein (X) and another antibody (B) binds to domain 4 of the same protein (X). I tried to use controls in my experiment. I used active and passive residues (derived from mutagenesis and structural interface) specific to antibody A and protein X (this is my positive control) docked them. The same thing I used for my antibody B which actually binds to domain 4 of that protein. I wanted to use it as a negative control. But HADDOCK does not seem to clearly differentiate between that? I get -110 HADDOCK score for my positive control and -98 for my negative control. Now I am stuck, if I cannot make a good statement why this is happening and then my test samples also wont be accepted right? Please advice me!

5

Remember docking scores are not binding affinities.

Any docking software will always give you an answer and not tell you: no it does not bind.

On one side, the good news is that you negative control has a worse score.

You could do the reverse experiment targeting domain 4