Recomputing MAICoS

[comcon1]

I started to recompute MAICoS on LUMI. There are some points.

1. recomputing is still quite unstable. Some systems are just not computed. But then you run analysis again and they are computed. Probably some small bugs with LUSTRE filesystems. But it's OK. I can progress even with this.
2. non-TPR systems. There are quite a lot of systems which doesn't have any topology. Just PDB. PDB doesn't have charges - everything breaks. Here we should immidiately turn off dielectric profiles, but all other profiles we MUST compute anyway.

[comcon1]

This block of code works very instable on Singularity/LUSTRE:

            os.system('rm foo.ndx')
            os.system(f'echo "a {last_atom}\nq" | {makendx_cmd} -f {tpr_name} '
                      f'-o foo.ndx')
            os.system("tail -n1 foo.ndx | awk '{print $NF}'")
            os.system('echo "[ centralAtom ]" >> foo.ndx')
            os.system("tail -n2 foo.ndx | head -n1 |  awk '{print $NF}' >> foo.ndx")

This is a part of output:

Group     3 (            POT) has    35 elements
Group     4 (            CLA) has    35 elements
Group     5 (           TIP3) has 40521 elements
Group     6 (             C1) has   300 elements
Select a group: 
-------------------------------------------------------
Program:     gmx trjconv, version 2025.2-conda_forge
Source file: src/gromacs/topology/index.cpp (line 873)

Fatal error:
Cannot read from input

For more information and tips for troubleshooting, please check the GROMACS
website at https://manual.gromacs.org/current/user-guide/run-time-errors.html
-------------------------------------------------------
Note that major changes are planned in future for trjconv, to improve usability and utility.
Select group for centering
Error: No such group 'centralAtom'
Selected 0: 'System'
Select group for output

Probably, it's a good point to switch os.system to subprocess / popen here. @batukav I saw you took already issue for subprocess, maybe you could start from this place? Then I could proceed recomputing...

[batukav]

I’ll look into this, but won’t happen before 05.08. Alexey Nesterenko ***@***.***> schrieb am Do. 24. Juli 2025 um 20:04:

…

This block of code works very instable on Singularity/LUSTRE: os.system('rm foo.ndx') os.system(f'echo "a {last_atom}\nq" | {makendx_cmd} -f {tpr_name} ' f'-o foo.ndx') os.system("tail -n1 foo.ndx | awk '{print $NF}'") os.system('echo "[ centralAtom ]" >> foo.ndx') os.system("tail -n2 foo.ndx | head -n1 | awk '{print $NF}' >> foo.ndx") This is a part of output: Group 3 ( POT) has 35 elements Group 4 ( CLA) has 35 elements Group 5 ( TIP3) has 40521 elements Group 6 ( C1) has 300 elements Select a group: ------------------------------------------------------- Program: gmx trjconv, version 2025.2-conda_forge Source file: src/gromacs/topology/index.cpp (line 873) Fatal error: Cannot read from input For more information and tips for troubleshooting, please check the GROMACS website at https://manual.gromacs.org/current/user-guide/run-time-errors.html ------------------------------------------------------- Note that major changes are planned in future for trjconv, to improve usability and utility. Select group for centering Error: No such group 'centralAtom' Selected 0: 'System' Select group for output Probably, it's a good point to switch os.system to subprocess / popen here. @batukav <https://github.com/batukav> I saw you took already issue for subprocess, maybe you could start from this place? Then I could proceed recomputing... — Reply to this email directly, view it on GitHub <#33 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB47SNLNJPRXJMPJO4IMEJT3KEN4RAVCNFSM6AAAAACCJLGRXCVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTGOBYGA4TSNI> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[comcon1]

@batukav Can you please prioritize this task? You can also wrap a train of commands into a single command like so

command = (
    "rm foo.ndx; "
    f"echo \"a {last_atom}\\nq\" | {makendx_cmd} -f {tpr_name} -o foo.ndx; "
    "tail -n1 foo.ndx | awk '{print $NF}'; "
    "echo \"[ centralAtom ]\" >> foo.ndx; "
    "tail -n2 foo.ndx | head -n1 | awk '{print $NF}' >> foo.ndx"
)

subprocess.run(command, shell=True)

[comcon1]

This block specifically has problems being run on singularity container. I think we could get rid of these trains of pipes here and work with foo.ndx using python interface. Also, I think, it's worth feeding make_ndx using popen pipes and not using bash pipes. I think, I will create an issue out of it.

[comcon1]

NMRLipids/FAIRMD_lipids#301