MSMH5 file

Using the native MSMH5 OpenMM reporter

import openmm as mm
from openmm import app
from openmm import unit

import molsysmt as msm
from molsysmt import pyunitwizard as puw

molsys = msm.convert('4POC')

molsys = msm.extract(molsys, selection='molecule_type=="protein"')

molsys = msm.build.add_missing_hydrogens(molsys)

molsys = msm.build.solvate(molsys)

topology = msm.convert(molsys, 'openmm.Topology')

coordinates = msm.get(molsys, coordinates=True)

Tip

Trajectory workflows can request selected atom data and system-level frame metadata together. For example, msm.get(molsys, element='atom', selection=atom_indices, coordinates=True, time=True) applies atom_indices only to the coordinates; time is evaluated at the system level. Results retain request order.

Tip

For a live OpenMM Simulation, msm.get(simulation, temperature=True, integrator=True, friction=True) retrieves metadata that the active integrator actually retains. Temperature and friction return None for integrators that do not define them. A force-field name is not inferred from an OpenMM System, because parameterization removes that unambiguous provenance.

positions = puw.convert(coordinates[0], to_form='openmm.unit')

forcefield = app.ForceField('amber14-all.xml', 'amber14/tip3p.xml')
system = forcefield.createSystem(topology, nonbondedMethod=app.PME,
        nonbondedCutoff=1.2*unit.nanometer, constraints=app.HBonds)
integrator = mm.LangevinIntegrator(300*unit.kelvin, 1.0/unit.picosecond, 2.0*unit.femtoseconds)
platform = mm.Platform.getPlatformByName("CPU")
simulation = app.Simulation(topology, system, integrator, platform)
simulation.context.setPositions(positions)
simulation.minimizeEnergy()
simulation.context.setVelocitiesToTemperature(300*unit.kelvin)

from molsysmt.third_party.openmm.reporters import H5MSMReporter, TQDMReporter

The reporter creates H5MSM 0.4 output. Its topology is stored as a native reference chemical state, while trajectory arrays remain chunk-friendly datasets. Files written by older MolSysMT releases remain valid 0.3 migration inputs.

tqdm_reporter = TQDMReporter(1000, 50000)
simulation.reporters.append(tqdm_reporter)

msmh5_reporter = H5MSMReporter('traj.msmh5', 1000, steps=50000, selection='all',
            topology=topology, time=True, box=True, coordinates=True, velocities=False,
            potentialEnergy=True, kineticEnergy=True, temperature=True,
            includeInitialContext=True, constantReportInterval=True,
            constantStepSize=True, constantBox=True,
            auto_close=True)
simulation.reporters.append(msmh5_reporter)

simulation.step(50000)

A first glance at the msmh5 file

aa = msm.convert('traj.msmh5', 'molsysmt.MSMH5FileHandler')

'topology' in aa.file

aa.file['topology']['atoms'].attrs['n_atoms']

msm.get('traj.msmh5', n_lipids=True)

Extract a new msmh5 file

When a long trajectory is processed in chunks, validate the requested frame range before starting expensive I/O. Public consumers such as msm.get(), msm.convert(), msm.extract(), and msm.Iterator raise molsysmt.ArgumentError immediately if a frame index is negative or outside the trajectory. Element selections and masks receive the same early validation, preventing a late backend-specific IndexError or a silent empty iteration after earlier work has already been performed.

# Bounded-memory ensemble descriptors
rg = msm.structure.get_radius_of_gyration('traj.msmh5', heavy_mode='force')
rmsf = msm.structure.get_rmsf('traj.msmh5', selection='backbone', heavy_mode='force')