Any easy way for Rigid-body MD in OpenMM?

[Seung Joong Kim]

Hi all,

I have a protein (~1200 residues) that consists of two domains (N: ~500 / C: ~700 residues each) and flexible linker (~20 residues) between N and C terminal domains. I would like to simulate how the N/C terminal domains can orient each other by moving the flexible linker only. I would treat each N/C terminal domain as a rigid body, so there should be no moving parts within the N/C terminal domains. The linker is the only moving part. Are there any easy way to simulate this? I mean, what is the function / expression to make a specific region as a rigid body and move the linker only?

In addition, I would like to add a (overall) size constraint in the system. I want to set a target "Radius of Gyration (Rg)" for the protein, so that by moving the flexible linker only, we can reach close to the target Rg (overall size of protein) over time. Is there any function/expression in python interface as well?

Thank you so much,
- Seung Joong Kim

[Seung Joong Kim]

< answer by Peter Eastman peastman@stanford.edu >

Unfortunately, there's no easy way to do that. Rigid body simulation involves very different techniques than all atom cartesian simulation, and OpenMM doesn't provide the tools to do it.

On the other hand, we actually have another product specifically designed for rigid body simulation of molecules: https://simtk.org/home/molmodel. I've cc'd Michael Sherman, who is the lead developer on MolModel. If you're interested in trying it, he can answer your questions about it. Also, MolModel can use OpenMM to accelerate its force calculations, so you still get to use your GPU

[Michael Sherman]

It is conceivable that Molmodel would do what you want, but by no means certain. It provides very limited access to OpenMM. I would suggest that you take a look at the Molmodel documentation https://simtk.org/home/molmodel to see whether it is even close to what you want. It is exclusively a C++ API; currently there are no Python wrappers.

If you find that Molmodel would be useful with some extensions and you are comfortable programming in C++, we would be delighted if you want to contribute to its development!

Regards,
Sherm

[Seung Joong Kim]

Thank you so much for reply. I'll read through the documentation of Molmodel, and post what I can do with it. Many thanks!

[David Minh]

I have a related question. I would like to do a docking-like calculation with a rigid receptor and flexible ligand, so that the ligand is sampled from the Boltzmann distribution for a given receptor conformation. So it isn't quite a rigid-body MD. Would OpenMM be able to do this type of calculation?

[Peter Eastman]

OpenMM 4.1 will let you immobilize any particle by just setting its mass to 0, so that would be an easy way to do that. We're hoping to have a beta out within the next few days. Would you like to try it?

Peter

[David Minh]

Yes, that would be great!

[Peter Eastman]

The beta is now available see this message for details.

Peter

[David Minh]

Good, I'll give it a shot.