Source code for CHARMM-GUI, other tools, and run optimization

GROMACS version: 2021.4
GROMACS modification: No

Hello all,

I’ll start by saying that I’m exceptionally new to GROMACS but I’m learning a bit more every day. Currently I’m generating the inputs for GROMACS through CHARMM-GUI and am modifying the .mdp files for my use case. I currently am able to run simulations but would like to add a variety of forcefields to my systems, such as SIRAH (SIRAH forcefield in GROMACS ~ SIRAH Forcefield) or AMBER99SBILD. In addition to that, I would also like to use the entirety of my GPU when running a production run.

So, my questions for you wonderful people.

  1. Is there any way to get the source code for CHARMM-GUI, and if not, are there any other tools like CHARMM-GUI where I could download the source code?
  2. If I can download the source code for CHARMM-GUI is there a way to add new forcefields to the program.
  3. Is there a way to increase the amount of GPU utilization while also keeping performance high?

There you have it, three questions in one post.

Looking forward to hearing your answers.

Best,
Kirtley

CHARMM-GUI is an incredibly complex software environment with may backend programs that require special licensing. If you want to integrate new force fields, you’ll have to suggest that to their developers and likely volunteer to be the one to do the testing of those new force fields.

GPU utilization is a function of how demanding your job is. There is no connection between GPU utilization and performance (that is, using 100% of the GPU does not mean your run will be faster than if it only uses 50% of the GPU).

Hmm, the forcefield portion is a bit unfortunate to hear.

Regarding GPU utilization. From what I’ve gathered there are portions of the MD that are better performed on CPU and some that are better performed on GPU. Is that true? If so, do you know what flags that should be sent to the gmx mdrun so that the software is distributed to run optimally, or is that built into it?

Most of the calculation can be offloaded to the GPU, just look at the available options in the mdrun help description.

Great tutorials by the way. The lysozyme in water and protein and ligand complex were great learning methods!