Question about using gromacs files as input

[obasekigit]

Good day, i am a newbie in using the a3fe package and would like to ask if i can use as input the output files generated from gromacs as input rather than the amber format.

[Roy-Haolin-Du]

Hi, thanks for your question!

You can quickly convert files if you’ve already installed A3FE,
Just simply open ipython to run:

import BioSimSpace as BSS
system = BSS.IO.readMolecules(['gromacs.gro','gromacs.top'])
BSS.IO.saveMolecules('output', system, ['prm7', 'rst7'])

Out[3]: 
['output.prm7', 'output.rst7']

Sorry, we currently hard-code the AMBER format, but GROMACS support will be added soon.

Enjoy a3fe! Let me know if you have more question.

[obasekigit]

Thank you
just some follow up questions
must the box size of the bound and free be the same size? does that matter?

[obasekigit]

i tried to run a quick ABS and a i got the following error about not having all the files needed but i have all the files from the looks of it attached below is the error and my directory

[Roy-Haolin-Du]

Thank you just some follow up questions must the box size of the bound and free be the same size? does that matter?

Hi, I think the box sizes for the bound and free systems are different. We determine the box size based on BioSimSpace.

a3fe/a3fe/run/system_prep.py

Lines 161 to 180 in e66afac

	# Determine the box size
	# Taken from https://github.com/michellab/BioSimSpaceTutorials/blob/main/01_introduction/02_molecular_setup.ipynb
	# Get the minimium and maximum coordinates of the bounding box that
	# minimally encloses the protein.
	print("Determining optimal rhombic dodecahedral box...")
	# Want to get box size based on complex/ ligand, exlcuding any crystallographic waters
	non_waters = [mol for mol in parameterised_system if mol.nAtoms() != 3] # type: ignore
	dry_system = _BSS._SireWrappers._system.System(non_waters) # type: ignore
	box_min, box_max = dry_system.getAxisAlignedBoundingBox()
	# Work out the box size from the difference in the coordinates.
	box_size = [y - x for x, y in zip(box_min, box_max)]
	# Add 15 A padding to the box size in each dimension.
	padding = 15 * _BSS.Units.Length.angstrom
	# Work out an appropriate box. This will used in each dimension to ensure
	# that the cutoff constraints are satisfied if the molecule rotates.
	box_length = max(box_size) + 2 * padding
	box, angles = _BSS.Box.rhombicDodecahedronHexagon(box_length)

which taken from https://github.com/OpenBioSim/biosimspace_tutorials/blob/main/01_introduction/02_molecular_setup.ipynb.

i tried to run a quick ABS and a i got the following error about not having all the files needed but i have all the files from the looks of it attached below is the error and my directory

For this issue, did you put the files in the right folder? They should be in ligand/input/*.prm7 and *.rst7, and you need to run IPython from the ligand folder.

Let me know if you have other questions.

Cheers~

[obasekigit]

Just to clarify both the bound and the free should be placed in the input folder? Or I should create the folders like seen in the tutorial where I have input for both bound and free?

[Roy-Haolin-Du]

Yes, both free_param and bound_param .prm7 and .rst7 files should put into the input folder. The bound andfreefolders will be generated automatically. You just need to make the input one.

Good luck!

[obasekigit]

Thanks for all the help @Roy-Haolin-Du . i guess i still have a question. the simulations starts now but i keep getting an error about gres specificaion (Error submitting job: sbatch: error: Batch job submission failed: Invalid generic resource (gres) specification). for background i am using my schools cluster and i dont have access to GPU now but i have CPU. how do i shut gres down?
below is my script im using for testing

import a3fe as a3
cfg = a3.SomdSystemPreparationConfig()
cfg.forcefields["ligand"] = "gaff2"
cfg.runtime_npt_unrestrained = 50 # ps
cfg.runtime_npt = 50 # ps
cfg.ensemble_equilibration_time = 100 # ps
calc = a3.Calculation(ensemble_size = 3)
slurm_config = a3.SlurmConfig(partition="batch", time="470:00:00", nodes=1, ntasks_per_node=24)
analysis_slurm_config = a3.SlurmConfig(partition="batch", time="00:10:00", nodes=1, ntasks_per_node=24)
calc = a3.Calculation(slurm_config=slurm_config, analysis_slurm_config=analysis_slurm_config)
calc.setup(sysprep_config = cfg)
calc.run(adaptive = False, runtime=0.1) # ns
calc.wait() # Wait for the simulations to finish
calc.set_equilibration_time(1) # Discard the first ns of simulation time
calc.analyse() # Fast analyses
calc.save()

[Roy-Haolin-Du]

Hi, thanks! A3FE through Slurm only supports GPU acceleration — there’s no CPU-only mode. This is because both GROMACS and SOMD, as well as the alchemical part calculations, all require GPU resources.
Try to find some available GPU resources for your runs.

As for your test script, all good, only
calc.set_equilibration_time(0.02) since your total runtime is 0.1 ns, discarding the first 1/5 (i.e., 0.02 ns) as equilibration is appropriate.

Hope it helps.

[obasekigit]

thanks for the response. is it possible to use google colab ?

[Roy-Haolin-Du]

I haven’t tried it myself, but I’m afraid it won’t work? Google Cloud probably doesn’t have a Slurm scheduling system either. If you really need to run some computational work, better to consider applying for computing resources through your relevant faculty.

[obasekigit]

oh i see. thanks again for the help

[obasekigit]

i was quickly able to hop on a GPU to test it but i got a sbatch: error: Batch job submission failed: Invalid generic resource (gres) specification. i known is because i specified the gres command on my submission script below but my clusters gpu does not use that command to access the gpu? any thought on how i can overcome this?

#!/bin/bash
#SBATCH --job-name=abs_V429
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=24
#SBATCH --partition=gpu
#SBATCH --gres=gpu:1
#SBATCH --time=48:00:00
#SBATCH --error=error.txt

[Roy-Haolin-Du]

You can override that directly,

import a3fe as a3

slurm_config = a3.SlurmConfig(
    partition="gpu",     # your GPU queue
    time="48:00:00",
    gres="",             # 👈 disable the --gres line
)

Detailed scripts on below: (change any key worsd using a3.SlurmConfig(xxx="xxx") )

a3fe/a3fe/configuration/slurm_config.py

Lines 20 to 30 in e66afac

	class SlurmConfig(_BaseModel):
	"""
	Pydantic model for holding a SLURM configuration.
	"""
	partition: str = _Field("default", description="SLURM partition to submit to.")
	time: str = _Field("24:00:00", description="Time limit for the SLURM job.")
	gres: str = _Field("gpu:1", description="Resources to request - normally one GPU.")
	nodes: int = _Field(1, ge=1)
	ntasks_per_node: int = _Field(1, ge=1)
	output: str = _Field(

[obasekigit]

i think is a problem with my schools GPU cluster as disabling gres does not work either

[obasekigit]

Hey @Roy-Haolin-Du just to circle back to this. my trace back error is from the _virtual_queue.py python script. is there a way to modify it so i dont "get Invalid generic resource (gres) specification" . jusing leaving gres blank seems not to work. attached below is the error

/home/obasekia/Downloads/fep/abs/a3fe/a3fe/run/_virtual_queue.py:276 in │
│ _submit_job │
│ │
│ 273 │ │ │ except Exception as e: │
│ 274 │ │ │ │ raise RuntimeError(f"Error submitting job: {process_ou │
│ 275 │ │ │
│ ❱ 276 │ │ return _submit_job_inner(job_command_list) │
│ 277 │ │
│ 278 │ def update(self) -> None: │
│ 279 │ │ """Remove jobs from the queue if they have finished, then move │
│ │
│ /home/obasekia/Downloads/fep/abs/a3fe/a3fe/run/_utils.py:100 in newfn │
│ │
│ 97 │ │ │ │ │ # Wait for specified time before trying again │
│ 98 │ │ │ │ │ _sleep(wait_time) │
│ 99 │ │ │ │
│ ❱ 100 │ │ │ return func(*args, **kwargs) │
│ 101 │ │ │
│ 102 │ │ return newfn │
│ 103 │
│ │
│ /home/obasekia/Downloads/fep/abs/a3fe/a3fe/run/_virtual_queue.py:274 in │
│ _submit_job_inner │
│ │
│ 271 │ │ │ │ slurm_job_id = int((process_output.split()[-1])) │
│ 272 │ │ │ │ return slurm_job_id │
│ 273 │ │ │ except Exception as e: │
│ ❱ 274 │ │ │ │ raise RuntimeError(f"Error submitting job: {process_ou │
│ 275 │ │ │
│ 276 │ │ return _submit_job_inner(job_command_list) │
RuntimeError: Error submitting job: sbatch: error: Batch job submission failed:
Invalid generic resource (gres) specification
151,1 97%