Interpreting Confusing Antibody Results


I have been using Haddock to dock a set of antibodies to antigens. For some antibodies the results are easy to interpret. However, other Haddock results contain conformations where the heavy and light chains of the antibody have been pulled apart (see cyan structure below).

Some of these antibodies are very weak binders in vitro; thus, I am unsure if these confusing Haddock results are a reflection of this observation or if there is something wrong with my Haddock setup. I have been running haddock with defaults settings aside from increasing the # of structures for rigid body docking, semi-flexible refinement, final refinement, and to analyze. Additionally, I have been defining the active residues of the antibody, while the antigen’s surface residues are defined as passive.

To summarize, in your (haddock team) professional opinion do you think the confusing Haddock results are a sign the antibody is a poor binder or should I adjust my Haddock run parameters (such as fixing the position of the antibody during it0) so that the results contain an in-tact antibody.


You must be running a local version of haddock. In this case your two chains are not covalently connected and can drift apart during the semi-flexible refinement (especially if there are clashes between them).

You need to define a few distance restraints between the two chains (e.g. between CA atoms, using the exact distance with 0 error bounds) and add those as unambiguous restraints. The server does that automatically. You can use for that for example the script from our haddock-tools


Dr. Bovin,

Sorry for not specificity how I am running Haddock but I am using the web sever. If the server defines these restraints automatically then this is very curious. Do you have any further suggestions?


Is your antibody defined as one single chain? This is required for Haddock

Where is the structure coming from?

The antibody is defined as one chain (A) and re-numbered to start from res number 1 using PDBtools. I’ve attached an example of a structure below just in case that will be useful to you.

The structures were generated using an alphafold colab notebook (Google Colab) The heavy and light chain peptide sequences were input individually and a multimer was generated. We were not able to run the relax step in this alphafold run; however, the resulting structure overlaps nicely with similar antibody structures determined via X-ray crystallography. .

Antibody_Example_PDB.pdb (139.5 KB)
In this example the heavy chain ends are res #114 and the light chain starts at res #115.

Then you most likely have clashes at the interface between the two chains that causes it to split…

Try running it first through the refinement interface of haddock

Running it through the refinement interface worked! Thank you for the suggestion I was not aware of this tool.


1 Like

Glad to hear :slight_smile: