Hi Atiya,

First, let me ask you a few questions:

1) Is 1bvy_solv.pdb the native PDB, or is it some model you made?

2) If you built the model using some web or local tool, which tool did you used?

3) Did you perform a disorder protein prediction (such as DISORPRED)?

4) Your protein has some ligants (according to PDB), are you using them, or did you deleted?

5) Fe in the structure sometimes couses problems, how did you manage the Fe binding?

Cheers

Bruno César Feltes

Thank you for your kind reply.

(1) 1bvy is the substrate free model which was then solvated in AMBER using ambertools.

(2) I used Amber tools 18 and then converted the topology to Gromacs format.

(3) No, I did not perform any such disorder prediction on my structure. Why do you think it is necessary? I think the structure after minimization is highly disordered and the topology needs a change but I am not sure.

(4) Which ligands are you talking about? Can you elaborate?

(5) I used the following tutorial to build the topology. In short, i used the Fe as a separate ion and then combined the HEM model with the protein one to generate the final topology.

Hope this helps! I appreciate your feedback and valuable time!

Perhaps I missed it, but what actual force field are you using? Not all of the force fields handle transition metals well. It's possible your force field is incompatible with the iron and when it tries to control it with Langevin dynamics, it makes the solvent collisions too violent on the protein. It will overpower other terms of the energy function and cause bonds, angles, etc. to distort from normal values, i.e. boom goes the protein.

Atiya Banerjee , the disorder prediction its not necessary, but could be useful for you to understand if, for some reason, your protein do have disordered regions that, although not clearly depicted in the PDB, could eventually arise anyway. It happened with me before. I worked with proteins that showed clear secondary structures where gone even before it reached 100ns of simulation time. Again, not necessary, but could be useful.

Moreover, according to PDB you have a FLAVIN MONONUCLEOTIDE and 1,2-ETHANEDIOL besides your heme. Those are the extra molecules I was talking about. Are they there, or did you deleted them?

Finally, Matt Coban is right. Force Field choice is of utmost importance here. I recently had a marathon of literature review because of a enzyme I was about to work that had a Fe, and it is extremely difficult to properly deal with some force fields for this kind of metal. Check that out, this can be your problem.

Bruno César Feltes ,

Thank you for the suggestion. Yes, I deleted the ligand before generating the topology.

Matt Coban ,

Force field is a challenge but I used AMBER 14ffsb. I do think that it is quite successful in case of heme containing proteins. Can you run a quick minimization and see what the issue is? If it is a topology problem, what would be the best way to tackle this? Any suggestions would be helpful.

Thanks,

Atiya

Atiya Banerjee ,

Unfortunately, my lab has no license for AMBER; I'm not sure how informative it would be if I am using entirely different software/setup. Did you run through the metal center parameter builder (MCPB) prior to minimization or just input the native PDB? If not, try using MCPB first. Aside from that, you could try a 12-6-4 Lennard-Jones nonbonded cationic dummy model instead of the bonded iron. If the nonbonded dummy model behaves well, then that effectively proves your issue is related to parameters involving the iron; if it also blows up, then your problem is more general. Hope that's helpful.

Matt Coban ,

I did run it through the MCPB program and it modeled accurately according to the tutorial. I do not think that the iron is causing problems. About AMBER, you dont need a license for using it in my opinion.

Here is the link. Use Option 2

http://ambermd.org/GetAmber.php#ambertools