[Dock-fans] Defining a bond between protein and ligand

Munoz.Ines imunoz at cnio.es
Fri Sep 4 09:41:33 PDT 2015

Dear all,

If I want to fix the bond between the selected atom in the inhibitor and an oxygen atom that belongs to the target protein…how do I define that into the docking script?..or should be ok by modeling into the PDB?
Thanks again.

From: William Joseph Allen <william.joseph.allen at gmail.com<mailto:william.joseph.allen at gmail.com>>
Date: Monday, July 13, 2015 4:57 PM
To: Inés Muñoz <imunoz at cnio.es<mailto:imunoz at cnio.es>>
Cc: "dock-fans at docking.org<mailto:dock-fans at docking.org>" <dock-fans at docking.org<mailto:dock-fans at docking.org>>, Trent Balius <tbalius at aol.com<mailto:tbalius at aol.com>>, Sudipto Mukherjee <sudipto.mukherjee at gmail.com<mailto:sudipto.mukherjee at gmail.com>>
Subject: Re: [Dock-fans] Defining a bond between protein and ligand

Hi Ines,

To my knowledge, DOCK does not support covalent bonds between protein and ligand. The best way that I can think of to model this is to run a fixed anchor docking (FAD) calculation. An example FAD input file can be found here:


One modification to this file would be to change "user_specified_anchor" to "yes". Then, specify an atom from the rigid segment that is covalently attached to the protein with the parameter "atom_in_anchor". Documentation for this step can be found here:


This assumes that the rigid segment of your ligand which is known to be covalently attached to the protein is already in the correct 3D spatial / rotational binding geometry with respect to the protein coordinates (e.g. from a crystal structure). Then, upon docking, the rest of the molecule will be flexibly built out from that starting point. If the covalently-bound rigid segment is not already in place, you will either have to model it in manually, or try to generate the correct binding pose by standard flexible docking.

I would also recommend turning the parameter "simplex_restraint_min" to "yes" so that the ligand does not slosh around too much in the binding site.


Finally, one problem that I can envision arising from this type of calculation: If the ligand and protein are typically covalently bound, but you model them as non-covalent, then your preparation steps may try to fill in that missing bond with hydrogen atoms on either side. Naturally, those hydrogen atoms would clash with each other in a very unfavorable way. You may need to manually replace those hydrogens with non-interacting, un-charged Dummy atoms, or something similar to that. Please take care to carefully think about how you approach this in case you need to justify it later.

Perhaps someone else will chime in on this if they think of a better approach. Let us know if you have any questions.

Joe Allen

On Mon, Jul 13, 2015 at 9:40 AM, Munoz.Ines <imunoz at cnio.es<mailto:imunoz at cnio.es>> wrote:

Could anyone send me an example of how to define and fix a bond between a protein residue atom and the atom of a compound, from which the interaction is well known, letting the rest of the compound molecule to adjust freely to the binding space during the docking proccess?

Thanks in advance and regards,

Dock-fans mailing list
Dock-fans at docking.org<mailto:Dock-fans at docking.org>

More information about the Dock-fans mailing list