Water Refinement HADDOCK 2.4

Currently I am trying to do water refinement of protein - ligand complexes I calculated with CYANA.
For this, I was wondering why HADDOCK web portal does not need any distance restraint inputs whereas for example the cns software accepts the restraints as input?
Thanks a lot in advance!

The refinement interface is a simple interface not allowing the definition of any restraints.

If you want to add restraints, you can do so via the regular submission interface following the following recipe to set up a refinement run:

But for the restraints section, provide your restraint (in CNS format) as unambiguous (or ambig if you want to allow for the default 50% random removal)

1 Like

Thanks a lot for the fast answer!
The param file + input section of the submit file section is temporarily disabled and forwards me to the normal submission.
Do I always need to provide 2 separate pdb files (ligand and protein) for the normal submission or does it also accept the complex structure?

You do need to provide those separately for the regular interface

1 Like

Hi there, docking and refinement are different protocols.

Docking: the goal is to “put molecules together” and for that you can use restraints to guide the spatial search and scoring.

Refinement: the interacting partners are expected to already be in a complex conformation and then both will be refined - without any spatial search, hence no need for restraints.

In our web platform I have made a interface that you can input your complex as one structure (each molecule in a different chain) - which follow the same steps of the advanced refinement (so you don’t need to click so many buttons :slight_smile:)

However if you have a specific case or if the restraint energy term is relevant to your study case, it’s recomended that you follow the advanced refinement guide and tweak it to your needs.


Thank you very much for your answers.
At the moment I have the problem, that the refined structures submitted by regular interface as described above vary a lot from the input structure. Furthermore, not all restraints are fulfilled anymore.

This raises the question for me if I am providing the wrong syntax. I always have distance restraints from my ligand to methyls of the protein and providing them like:

assign (resid 500 and segid B and name H6) (resid 56 and segid A and name HD2#) 3.22 0.32 0.32
assign (resid 500 and segid B and (name H2 or name H3)) (resid 56 and segid A and name HD1#) 3.82 0.38 0.38

In the second case, the chemical shift of the ligand for H2 and H3 are the same. In CYANA I worked this around with a pseudo atom Q23.

Is this syntax correct for distance restraints in the .tbl file?

Thank you very much in advance!

Can you share a run ID?

Did you keep all hydrogens?

Did you turn off (set the number of steps to 0) for all flexible refinement stages?

In principle water refinement only should not change the structures so much

One run ID would be for example: 253719
If no error or misunderstanding happened, I followed those steps.

Hi Matthias,

Tooks a look at your run.

In principle the settings are fine (except you can reduce the sampling for refinement to 20 for all stages).

Now the problem comes because our generated ligand topo only contains polar hydrogens.
Which means none of your restraints were applied.
Unfortunately, the PRODRG version we have at hand does not support fully protonated ligands.

The only solution would be to apply the distance restraints to the heavy atoms of the ligand (with a correction to the distance).


1 Like

Hi Alexandre,

Thank you for your answer and help. I will try it with the carbon - methyl distances.