I am new to the HADDOCK2.4 web portal and am attempting to model an icosahedral viral capsid composed of 60 units (PDB ID: 1LP3). I have identified the interface amino acid positions (active residues) for the 2-fold, 3-fold, and 5-fold symmetry axes using 1LP3.pdb. When I attempted to reproduce a homodimer around the 2-fold axis and a homodimer or homotrimer around the 5-fold axis, HADDOCK2.4 successfully predicted the known structures of multimers. However, my attempts to model a homodimer or homotrimer around the 3-fold axis have been unsuccessful. All the predicted structures show a much looser association than the known structure, despite providing a large set of interface amino acids (199 amino acid positions as active residues) to the model.
I am uncertain why this approach works for the 2-fold and 5-fold multimers but not for the 3-fold multimers. It is possible that the protein-protein interface in the 3-fold multimers is highly interdigitated (like the interlocking of fingers in a handshake), making docking simulations particularly challenging. This might require more relaxation in the “it1” step.
I would greatly appreciate any assistance in resolving this issue to successfully model the multimerization of viral protein units around the 3-fold symmetry axis. If needed, I am happy to share the information about the active amino acid residues I used for the docking.
I am uncertain why this approach works for the 2-fold and 5-fold multimers but not for the 3-fold multimers. It is possible that the protein-protein interface in the 3-fold multimers is highly interdigitated (like the interlocking of fingers in a handshake), making docking simulations particularly challenging. This might require more relaxation in the “it1” step.
That could well be the case. It1 won’t solve your problem though.
If you do expect indeed more interdigitation, what you could try is to reduce the intermolecular interactions (vdw) during the rigid body docking phase (the inter_rigid parameter in run.cns - on the server in the "Scaling of intermolecular interactions for rigid body EM” under the Energy and interaction parameter menu in the docking tab).
Set it to 0.001 and also set the weight of the vdw energy term to 0 for rigid body.
Hi Dr. Bonvin, thank you very much for your suggestion. I followed your directions to model a homodimer around the 3-fold symmetry axis. By “homodimer,” I mean that I focused on two subunit proteins that form a part of the 3-fold symmetry axis of the icosahedral viral capsid. I used the following customized settings: under “Energy and interaction parameters,” I changed the “Scaling of intermolecular interactions for rigid body EM” from 1.0 to 0.001, and under “Scoring parameters,” I adjusted Evdw1 under Evdw (underlined) from 0.01 to 0. This setting generated homodimer docking models around the 3-fold symmetry axis that showed a greater degree of interdigitation between the two subunits, but it was not as pronounced as that in the known homodimer structure.
Would you mind providing me more guidance on how to achieve a simulation that results in a greater degree of interdigitation more closely resembling the known structure? For this docking simulation, I designated all interface amino acids as active residues (235 residues out of a total of 519, with 110 for Molecule A and 125 for Molecule B). I also did the simulation using 199 interface residues as active residues and using the same parameter setting, showing lesser degree of interdigitation.