Error with gmx_MMPBSA

[Zaiabari]

I've used Gromacs for MD simulation of Ligand(PCB)-Protein(4f3l). There are 30 ligands with one protein. I generated topology by CHARMM_GUI. I'm going to calculate the free energy of the structure by gmx_MMPBSA. I followed the instruction for the structures with LP pseudo-atoms. However I face a problem here:
gmx_MMPBSA -O -i mmpbsa.in -cs str_noLP.pdb -ci index.ndx -cg 1 17 -ct com_traj.xtc -cp topol.top -o FINAL_RESULTS_MMPBSA.dat -eo FINAL_RESULTS_MMPBSA.csv
[gapsoo-Precision-7920-Tower:15647] shmem: mmap: an error occurred while determining whether or not /tmp/ompi.gapsoo-Precision-7920-Tower.1000/jf.0/3927703552/shared_mem_cuda_pool.gapsoo-Precision-7920-Tower could be created.
[gapsoo-Precision-7920-Tower:15647] create_and_attach: unable to create shared memory BTL coordinating structure :: size 134217728
[INFO ] Starting gmx_MMPBSA 1.6.4
[INFO ] Command-line
gmx_MMPBSA -O -i mmpbsa.in -cs str_noLP.pdb -ci index.ndx -cg 1 17 -ct com_traj.xtc -cp topol.top -o FINAL_RESULTS_MMPBSA.dat -eo FINAL_RESULTS_MMPBSA.csv

[INFO ] Checking mmpbsa.in input file...
[INFO ] Checking mmpbsa.in input file...Done.

[INFO ] Checking external programs...
[INFO ] cpptraj found! Using /home/gapsoo/anaconda3/envs/MMPBSA/bin/cpptraj
[INFO ] tleap found! Using /home/gapsoo/anaconda3/envs/MMPBSA/bin/tleap
[INFO ] parmchk2 found! Using /home/gapsoo/anaconda3/envs/MMPBSA/bin/parmchk2
[INFO ] sander found! Using /home/gapsoo/anaconda3/envs/MMPBSA/bin/sander
[INFO ] Using GROMACS version > 5.x.x!
[INFO ] gmx found! Using /usr/local/gromacs/bin/gmx
[INFO ] Checking external programs...Done.

[INFO ] Building AMBER topologies from GROMACS files...
[INFO ] Get PDB files from GROMACS structures files...
[INFO ] Making gmx_MMPBSA index for complex...
[INFO ] Normal Complex: Saving group Protein_lig (1_17) in _GMXMMPBSA_COM_index.ndx file as _GMXMMPBSA_COM.pdb
[ERROR ] MMPBSA_Error

/usr/local/gromacs/bin/gmx editconf failed when querying com_traj.xtc

Check the gmx_MMPBSA.log file to report the problem.
File "/home/gapsoo/anaconda3/envs/MMPBSA/bin/gmx_MMPBSA", line 8, in
sys.exit(gmxmmpbsa())
File "/home/gapsoo/anaconda3/envs/MMPBSA/lib/python3.10/site-packages/GMXMMPBSA/app.py", line 98, in gmxmmpbsa
app.make_prmtops()
File "/home/gapsoo/anaconda3/envs/MMPBSA/lib/python3.10/site-packages/GMXMMPBSA/main.py", line 681, in make_prmtops
self.FILES.mutant_receptor_prmtop, self.FILES.mutant_ligand_prmtop) = maketop.buildTopology()
File "/home/gapsoo/anaconda3/envs/MMPBSA/lib/python3.10/site-packages/GMXMMPBSA/make_top.py", line 123, in buildTopology
self.gmx2pdb()
File "/home/gapsoo/anaconda3/envs/MMPBSA/lib/python3.10/site-packages/GMXMMPBSA/make_top.py", line 320, in gmx2pdb
GMXMMPBSA_ERROR('%s failed when querying %s' % (' '.join(comprog), self.FILES.complex_trajs[0]))
File "/home/gapsoo/anaconda3/envs/MMPBSA/lib/python3.10/site-packages/GMXMMPBSA/exceptions.py", line 171, in init
raise exc('\n\n' + msg + '\n\nCheck the gmx_MMPBSA.log file to report the problem.')
MMPBSA_Error:

/usr/local/gromacs/bin/gmx editconf failed when querying com_traj.xtc

Check the gmx_MMPBSA.log file to report the problem.
Exiting. All files have been retained.

[Valdes-Tresanco-MS]

Please, attach the gmx_MMPBSA.log file to check the error log

[Zaiabari]

Hi There,
The log file is attached.
Thanks
Ali
gmx_MMPBSA.log

[Zaiabari]

Hi there, The log file is attached. Thanks Ali

…

On Tue, Jan 7, 2025 at 3:03 PM Mario Sergio Valdés Tresanco < ***@***.***> wrote: Please, attach the gmx_MMPBSA.log file to check the error log — Reply to this email directly, view it on GitHub <#559 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BBFWKEAOT2MY3UNSD5F6FOT2JQ6JVAVCNFSM6AAAAABUYJ637GVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTCNZWGU3TEMA> . You are receiving this because you authored the thread.Message ID: <Valdes-Tresanco-MS/gmx_MMPBSA/repo-discussions/559/comments/11765720@ github.com>

[Valdes-Tresanco-MS]

It seems like a gromacs error similar to this one https://gromacs.bioexcel.eu/t/trjconv-gives-error-munmap-chunk-invalid-pointer/3166. I suspect it is related to the maximum number of groups that can be processed. According to the documentation (https://manual.gromacs.org/2024.2/reference-manual/algorithms/group-concept.html), there is a limit to some concepts and tools. I am not sure if this case is related, but it could be that trjconv fails when trying to read an index with so many groups. Did you run your simulation with this version of gromacs? I recommend, not only to check if it is the problem but also for readability, that once you remove the LP atoms, you remove all the groups that were created that are not necessary. It is not the same as selecting group 20 instead of the 800.
Are you sure that the selected group is correct? Here you are only selecting LIG_CL1_10122 (group 17)
I find it very interesting because your ligand has 780 atoms. Are all 30 ligands bound simultaneously? I thought there were 30 different simulations.

[Zaiabari]

I considered 30 ligands and one protein and ran the simulation using Gromacs 2024.2. I used Charmm-GUI to build the topology. All ligands are not bound simultaneously.

[Zaiabari]

I've got stuck though I followed all your instructions. Still I see an error (gmx editconf failed when querying com_traj.xtc). So, I send all files. Please check it out. FYI I have a receptor (Protein #4f3l) and 30 identical ligands (PCB 77; containing halogen (Cl)). I generated the topology using Charmm-Gui and performed the MD simulation by Gromacs. I also edited the LIG.itp file in toppar directory and removed all LP atoms plus exclusions and possible pairs related to LP atoms. As well, I followed your instruction to splitat the Ligand (30 ligands contain 120 LP atoms that I groped all 120 LPs (group number 797) and using 13&!797 made a new group for ligand without LPs (group #798 named lig). Then I used 1|798 (Protein-lig) followed by "del 17-797" . Finally I have 18 groups ... the rest of the procedure performed according to your instruction. I attached the Zip file including all necessary files.
toppar.zip
tpr top index in.zip
the rest of the files are attached with the next comment

[Zaiabari]

MD.xtc file through Google derive links
https://drive.google.com/file/d/1qsVw0YvZcvVl7NKx43Y4OkCm7PLBY58h/view?usp=sharing

[Zaiabari]

com_traj.xtc file
Uploading com_traj.xtc file.zip…

[Valdes-Tresanco-MS]

I finally found the problem. I will point out some points that I think are important.

1. Without wanting to belittle your work, I don't see any point in the simulation, much less calculating the energy in this case. The trajectory shows that the ligands were all aggregated (image). Any information here will probably lead to erroneous conclusions. I recommend you study the limitations and challenges of fragment simulation, which is essentially what it is.
2. The ligand has some problems. After two hours of testing and reviewing your files, I concluded that the problem is in your ligand. For some reason, editconf doesn't want to deal with anything that has to do with your ligand when a PDB file is used as the structure (necessary in this case). For any selected group that doesn't contain the ligand, it saves it correctly. In any case, where the ligand is involved, it saves the PDB, but the program closes unexpectedly. This makes gmx_MMPBSA think that it ends in error and does not continue. To rule out that it was a direct error from gromacs I checked if a similar error had been reported and I didn't find it. I also tested with several systems and it only fails on yours.
3. Considering both points, it is better to try with a single ligand. If it fails, we will investigate the cause. Unfortunately, we don't have much time to resolve issues.
   If you have any questions, I will gladly answer them if possible.

[Zaiabari]

This is a problem that requires consideration of multiple ligands, as the number of ligands interacting with the PAS-B region of the protein is crucial for comparison with other types of ligands.

[Zaiabari]

I tried it with one ligand, and it worked. However, as I mentioned, your code fails when there are multiple ligands. I'm trying to use another forcefield (AMBER) over topology generation and check if your code works for multiple ligands

[Valdes-Tresanco-MS]

gmx_MMPBSA works for complex systems—for example, Protein+DNA+RNA+Ions.
Note that the error is not related to gmx_MMPBSA but to gromacs. The fault is in editconf, if you manage to make editconf generate the pdb files for the complex, receptor, and ligand, gmx_MMPBSA will work fine, it is more than tested. As long as editconf terminates abruptly with a Segmentation Fault, there is no way for gmx_MMPBSA to do anything about it because it is not related to it.

[Zaiabari]

Do you think it works if I manage to generate the same complex without LPs? I mean the complex that is made for MD simulation and not that one is made over your instruction (by removing the generated LPs)

[Valdes-Tresanco-MS]

I don't know. As I said, the problem is actually with editconf. For some reason, it doesn't process the tpr or pdb where it has its ligand. If you test it, you'll see that even though it saves the pdb, editconf closes abruptly with a Segmentation Fault message. This tells gmx_MMPBSA that it didn't finish properly, even though it generated the pdb file. That is, the tests you do should focus on reproducing the steps that gmx_MMPBSA does with gromacs (you can check them in the gmx_MMPBSA.log file in the tests). Once you make editconf work correctly then gmx_MMPBSA shouldn't have any problems.