I want to compare WT vs mutant protein KD to DNA (and possibly inhibitors)... I have questions

User discussions
1

GROMACS version: 2025.2
GROMACS modification: No

I am running MD simulations using structure predictions generated by Boltz-2 (AlphaFold3) of a homo- and hetero- dimeric transcription factor that is a known oncogene with 82 known gain of function mutations. it is a large protein thus creating a large solvation box with a total of 700,000 elements. (protein contains 24620 elements, the rest is solvent) what is the shortest time I can run a simulation to obtain a relative Kd value of the wild type protein against its mutants?

I understand that usually these simulations need to run on the hundred- nanosecond to microsecond scale, but is there a work around since I have so many mutants?

I was able to run a 100 ps simulation on my laptop with processing from pdb taking about 5 hours. I have plans to move the simulation to AWS.

Any advice would be greatly appreciated.

Thank you!

Aaron

2

Hi @schulab ,

How do you want to calculate the Kd? 700K atoms is quite a large systems, to be honest I would go with a few tens (if not hundreds) of ns just to fully equilibrate the box.

4

Hi obZehn,

I know i want to calculate free energy using gmx_MMPBSA (or if another method would be better)

thank you for the quick reply

5

I do not have enough experience to comment on gmx_MMPBSA, but probably it still requires and ensemble of frames, and I would argue that 100 ps is not enough with such a large system. There are a myriad of methods to calculate free energies, it really depends on what you want to do specifically. Anyway, I wouldn’t rely on simulations long 100 ps for anything that might have some major degree of freedom to relax, like a large protein!