Error during the last analysis stage of water refined structures


I am using Haddock 2.2 locally to dock protein-RNA complex. The starting protein structure is obtained from the PDB and contains two water molecules. The whole docking procedure goes perfectly fine except at the very last stage of analysis after water refinement. Some of the water refined structures are having really huge energies and Haddock scores, due to “explosion” of the protein, actually it becomes a very long extended chain. Also, H-bond and hydrophobic contact analysis fails, the scripts are trying to analysis the protons of the starting water molecules, but naturally those are no longer existing after water refinement, they are deleted along with the rest of water molecules added during the refinement step. Also, the fit structures and the average structures of this water refinement step are inheriting the “exploded” protein structure. As a result, clustering does not provide any helpful results.
Alternatively, deleting these two water molecules from the beginning would resolve all the errors. However, the docked structures are totally different from the ones containing water molecules. It seems that the protein/RNA has more flexibility during it0, and the provided distances are not restraining the docking properly as it is doing in presence of the water molecules. For example, I get very defined structure satisfying my intermolecular restraints, and most of it1 structures are converging into one cluster only. Removing the water molecules results in more “random” structures, only 60%of them are clustered, and now covering more than 11 clustered rather than one!

The waters are defined as “HOH” residues in the initial protein structure, and I assume they would not change a lot of the protein structure.
It would be extremely appreciated if you can help me resolve this problem.

Best regards,

Difficult to say what is happening, but seems to me a problem with your starting structures or your setup.

Did you try running it on our web server to see if the problem persist?

Thanks a lot for the quick reply!
Unfortunately, I am using dummy atoms during these runs to replace PRE tags, this is not available on the Webserver.
I will try and start two identical runs locally and on the server using AIR only and without dummy atoms to see if the two are producing similar results or not.
Will report back once I have the results.

Best regards,