MD simulation of a dimer protein -dimer ligand complex

GROMACS version:2020.1-Ubuntu-2020.1-1
GROMACS modification: No
I am trying to do an MD simulation of a protein-ligand complex where both the protein and ligand are dimers. I am following the tutorial here.

For my ligand, I saved it in a cmp.pdb file (cAMP is my ligand). I added H-atoms using Avogadro, as shown in the tutorial. In the tutorial, we are instructed to edit the mol2 file to fix the residue names and numbers such that they are all the same. Since I have two molecules I made sure that all atoms and H atoms for cAMP1 were labelled with the same residue names and numbers and for cAMP 2 with same residue names and numbers. Having done this, I ran the perl script as shown in the tutorial.
Here is the final mol2 file Pastebin

Next, I went on to CGenFF and submitted my mol2 file. I had the following error in the error log.
“readmol2 warning: non-unique atoms were renamed. Now processing molecule CMP …”

Here is the output file Pastebin.

I am getting terrible penalty scores and figure I must be going wrong somewhere. Where did I go wrong? How should I approach the task of doing MD simulations for a protein dimer ligand dimer complex?

Your .mol2 file should only contain one copy of cAMP. You’re simulating two copies of the same thing, so the topology is identical. Just specify 2 copies of cAMP in [molecules] in your system .top and that’s all you have to do.

1 Like

Oh I understand now. Thanks a ton! And thanks a ton for the detailed tutorials as well

I extracted a single copy of my Ligand, and generated a mol2 file after adding Hydrogens using Avogadro. I processed the file using the perl_sort file, and also made more edits following the tutorial.
I loaded this into CGenFF, and got a str file output.
The problem is, I am still getting very high penalty score values, and am unsure if I should proceed with the simulation.
Here is a copy of the CGenFF output.

I notice very high charge penalty values for the P atom, but I can’t just remove this atom and proceed, right? What optimization/validation steps could I do to proceed with the analysis?
Here is my PDB file with just one cAMP molecule
Here is my processed mol2 file which I feed to the CGenFF server.

You should refer to the 2010 JCC CGenFF paper, which is itself a tutorial for how to parametrize molecules with the CHARMM force field and how to adjust parameters. The charge penalties do not necessarily mean the parameters are wrong, just that there is low confidence in the prediction. Note that you have symmetry that is not being properly accounted for (e.g. the O1P and O2P charges should be the same, likely O5’ and O3’ should be, also). This simplifies fitting a bit, and you do not need to touch the adenine base portion of the molecule.

1 Like

Hi, I think that the cAMP ligand is already parametrized, and just wanted to confirm that.

I went over to the Mackerall lab website, and downloaded the latest additive force field files toppar_c36_jul22.tgz. I proceeded to the stream sub-directory and opened the toppar_all36_na_nad_ppi str file. Here, I found the CMP molecule parametrized.

Here is the output of the CMP molecule; I verified the structure to be same as that of my own mol2 file. The only difference I can spot in the str file of the above link and that of the str file with high penalty scores I generated here is that the CMP molecule in the first one has a charge -2 (RESI CMP -2.00), while the one I generated has a charge of 0 (RESI CMP 0.000)
Can I proceed with the str file I found?

Edit: The default str file does not have DIHEDRALS.

CMP (as noted in the residue definition) is cytosine monophosphate. It is not the same as cAMP (cyclic adenosine monophosphate), which is not in the force field by default.

Your original .mol2 file has an incorrect protonation state. The phosphate group should be deprotonated and carrying a -1 charge.

When there are no additional parameters defined in a stream file, it means all the necessary parameters already exist in the force field. This is not diagnostic of anything, really.

Oh I see, my bad. I’ll work on using FFParam to optimize the ligand parametrization.