Error with multiple trajectory MMPBSA

[Ildar]

Hello, dear developers!

I have a problem when I try to do multiple trajectories MMPBSA with decomposition for the protein-ligand system modeled with the CHARMM force field. Single trajectory MMPBSA worked fine for the system.

 The error is quite repetitive and I attach the full error file along with input and run files.  File gmx_MMPBSA.log is empty, program crashes at the beginning.

 In essence, the error is:

    

from mpi4py import MPI

ImportError: dynamic module does not define module export function (PyInit_MPI)

[Mario Sergio Valdes]

Do you mean serial mode or Single Trajectory protocol? I assume you mean the first one... The first one runs gmx_MMPBSA in a single thread, doesn't use mpi4py, and can run both Single and Multiple Trajectory protocols. The second one uses the components from the complex without the need for separate receptor and ligand trajectories. Since the problem is importing mpi4py, I assume it is a compatibility problem.
Please check your mpirun version (mpirun -info) and the mpi4py version (conda list). Remember to activate the conda environment you created for gmx_MMPBSA.
In most cases, HPC administrators encourage users to use the MPI they compiled since it is optimized to use HPC resources. This is recommended and extremely useful for programs compiled with MPI. Given the architecture of gmx_MMPBSA, where the computation and the trajectory are split into as many parts as CPUs, the MPI installed with conda should suffice, and you won't see any performance issues. This ensures compatibility between mpirun and mpi4py.

[Ildar]

I meant Multiple Trajectory protocol. 

The output of conda list did reveal to me that mpich is already in install in conda env and the problem was caused by the conflict with HPC's mpich package, so I just do not use last anymore. My initial task is solved, and thank you so much, Mario!

But now I have another issue and I'm not sure is it solvable. 

Short description of my research. I study an enzyme that can bind 4 ADP molecules at effector sites, and this causes allosteric changes in structure. I applied Single Trajectory Protocol for it and it did well. Now I want to apply Multiple Trajectory Protocol to consider the difference between apo and the bound structure of the protein in delta energy calculations, but can't do it. Seems like the ligand's files need a total match between each other in an atom count, but I can't manage to make a .mol2 file for the system of 4 ADP molecules.

So, my question is it possible now to apply Multiple Trajectory Protocol for the system with binding 4 ligands, all at once? Am I doing something wrong?

Kind regards,

Ildar

суббота, 20 мая 2023 г. в 22:26:21 UTC+8, mariosergi...@gmail.com:

[Mario Sergio Valdes]

Yes, you can do it using the -cp topol.top option (which does not require the -lm mol2 file). Please, note that you can compute the energy only for one ligand at a time. If you want to keep the other ligands bound, make sure the receptor group contains the protein and the rest of the ligands (in this case, 3). You can check the consistency using gmx editconf with the proper index.ndx

[Ildar]

So, I can't do calculations in a way like dG = G(enzyme + 4 ligands) - G(apoenzyme) - G(4 ligands)?

понедельник, 29 мая 2023 г. в 23:16:35 UTC+8, mariosergi...@gmail.com:

[Mario Sergio Valdes]

Yes. You must create a group with the receptor and another with the four ligands.