Using the pmda analysis classes

Pre-defined analysis classes follow a common pattern. They require one or more MDAnalysis AtomGroup instances as input (which one obtains from a MDAnalysis Universe, typically by performing a selection). They also typically require a number of parameters that are provided as keyword arguments:

class ParallelAnalysisClass(atomgroup, [atomgroup, [..., ]]**kwargs)

set up the parallel analysis

run(n_jobs=- 1)

perform parallel analysis; see Parallelization for explanation of the arguments


stores the results (name and content depends on the analysis that is being performed)

The instance of the parallel analysis class contains a run() method. Calling run() performs the parallel analysis. Using keyword arguments, one can set the level of parallelization.

After the run has been completed, results are stored in an attribute of the class, such as results. The actual name of the attribute is defined by the specific analysis class. The format and content is also dependent on the analysis; for example, the RMSD analysis class pmda.rms.RMSD produces a numpy array of the RMSD time series.

Example: Parallel RMSD

In order to use the parallel RMSD calculator pmda.rms.RMSD, import the module, set up two AtomGroup instances, instantiate the class and directly use its run() method:

import MDAnalysis as mda
from pmda import rms

u = mda.Universe(top, traj)
ref = mda.Universe(top, traj)

rmsd_ana = rms.RMSD(u.atoms, ref.atoms).run(n_jobs=-1)


The resulting rmsd_ana instance contains the array of the RMSD time series in the attribute rmsd.