Protein protein docking

I run a protein peptide docking on Grue interface. I only chose the peptide(10AA) flex. but i found the following error:

Flexibility and restraints are two different things.
Making the peptide completely flexible will just impact (and increase) its capacity to undergo conformational changes during the simulated annealing stage (it1) and refinement step (water) of HADDOCK. However, you need restraints to guide the docking and more particularly the first step of HADDOCK (it0).

Did you provide any ambiguous or unambiguous restraints? Or did you turn on the CM mode (Center-of-mass docking)?

No, I did not. I used default setting.

Then you need to provide restraints. If you don’t have any experimental data that you can convert into distance restraints and give as unambiguous or ambiguous restraints, you can provide active/passive residues that you think are important for the interaction between the protein and the peptide.

Finally, if you really have no information, you can try an ab initio docking by turning on the Center of mass restraints parameter under the “Distance restraints” section of HADDOCK Guru interface.

1 Like

I have a question
I did docking with haddock
1.Beacause i didn’t have any information about active residue, i select all residue in two protein as active residue 2.because one of two protein don’t have good 3-D model, i use full flexible for this one,3- and I use 3 type of docking in guru interference simultaneously
Is it correct?


I’ll try to answer point by point:

  1. Selecting all residues is overkilling because the buried residues will certainly not interact with the other molecule. We usually advice to take all residues that are solvent accessible as active residues. Passiev residues can then be guessed automatically by HADDOCK if you check the option.
  2. Did you try modelling it via a template? Use HHPred for instance to try to find homologs and then model it with MODELLER. Or you can go for automated web servers like RaptorX or Robetta. HADDOCK full flexibility might not be the best approach to try folding your protein and dock it at the same time.
  3. I did not understand this part, what are the 3 types of docking you are mentioning?

Good luck,


1 Like

thank you for your answer.
first, I wanted to model protein 1 with the modeler, but I didn’t find any template.
so i decided to predict two protein 3-D with i tasser.
i know one of the protein high likely dock with himself.
now I want to dock this protein with another protein.
I mean three types of docking ab-initio containing the center of mass, Random patches, and Surface contact restraints.

Of course, I did docking with the guru interface and the active residues were proposed with i-tasser, but rmsd was 11.7. the result is uploaded
Capture (1)

This RMSD value indicates the average distance between the best 4 models of this cluster (#2) and the best scoring model generated by HADDOCK. So this is only an information about how much the best 4 models of this cluster deviate compare to the best scoring model. This is clearly not an information about how close those models are from the real complex since we do not know the answer. You should look at the representatives of the top 5 clusters for instance and investigate how different they are. Is there some convergence towards a particular solution (a cluster with many models for instance)? etc.

1 Like

In your opinion, can i write this score in my article?
I heared that rmsd above 2 is invalid. Is it correct?
In your opinion can I follow this protocol?

  1. Insert solvent residue that proposed i tasser as active residue?
    2.increase sample number 1000-400-400
    3.analysis that you say

RMSDs reported by HADDOCK have no meaning in terms of quality of models. They only tell you were the current cluster is with respect to the best model.

Considering the many unknown in you modelling, starting with the quality of the homology models, I wouldn’t have much trust in any of your results unless you could validated your models (e.g. by proposing and testing mutations at the interface of your complex).

You can always report numbers, but they might not have any meaning / value…

1 Like

Sorry, I ask a lot of questions.
Is the path, I wrote above, true for Docking?
That is, i do the same modeling and docking that I mentioned above to check the mutated protein form of one of them?
In which part of the haddock result I can find the interaction energy of each docking?

Which path do you refer to?

The HADDOCK score and its components are given on the result page of the server.
Further we do have more info about analysis in general on the HADDOCK software page


And more specifically:

The analysis described here would require downloading the full tar archive of a run.

1 Like

1.Insert solvent residue that proposed i tasser? or Insert Ligand Binding Site Residues that proposed i tasser?

2.increase sample number 1000-400-400

I am afraid I don’t understand what you mean under point 1)…
If you mean to define all solvent accessible residues as active then it’s a bad idea… Better use CM restraints.

Or switch to using a true ab-initio docking server like CLUSPRO for example.

1 Like

thank you for answer
my friend did one docking for me
she writed to me in protocol:“in order to docking we focuced on our data driven from docking approach,ie haddock using sequencial docking method…
since molecular dynamic isnot really suited for docking purpose because sampling in docking would be extremely computational expensive,so haddock could be used to refine model”
now i dont access to my friend.
from these sentences i understand that she created air file. but i don’t know how did she determine active residue? and what is the sequential docking method?
in another sentense she said that “molecular dynamic isnot really suited for docking …”,Is molecular dynamic different from refine model by haddock ?

Well, I don’t have access to your friend either… How can I answer those questions???

Excuse me for my question

Hi, I am Ilgaz.

I am applying protein protein docking via HADDOCK server. I am docking gp350 protein to cr2 and in each docking process, I apply some different mutations to gp350 before docking. I have 4 mutations as total and one of them is control mutation, whose binding affinity to cr2 increases after related mutations (experimental data). However, when I applied solvated docking with solvent shell, its binding energy becomes positive. On the other hand, binding energy of other mutations also becomes positive after same method (which is okay for me as experimental data supports these results) but I could not be sure whether these binding energy results are true after I saw an unrequired binding energy result for control mutation.

On the other hand, when I don’t add solvent shell and only apply solvated docking, binding energy of control mutation turns into what I want. However, I could not get binding poses as good as the ones I got from the solvated docking applied with solvent shell (By the way, I have the binding position of cr2-wt_gp350 docking, so, I have reference binding pose for comparison.).

Could you please help me about this situation? Why binding energy results of solvated docking with solvent shell method are positive?

I am looking forward for your answers, thank you.

I would not suggest using solvated docking. (Do you mean by that the option offered by the server?)
What kind of data are you giving to HADDOCK? I.e. what are your input data and parameter setup?

Do you have a structure of the wild type complex?

If this is the case, then I suggest rather running only a refinement and not full docking, introducing the mutations in your input PDB files.

For that only edit the name of the amino-acid you want to mutate, don’t bother changing/deleting atoms.

Also remember that there is in general poor correlations between binding affinity and docking scores as we demonstrated in the past.

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)

And check also: