MDTraj is a python library that allows users to manipulate molecular dynamics (MD) trajectories and perform a variety of analyses, including fast RMSD, solvent accessible surface area, hydrogen bonding, etc. A highlight of MDTraj is the wide variety of molecular dynamics trajectory file formats which are supported, including RCSB pdb, GROMACS xtc and trr, CHARMM / NAMD dcd, AMBER binpos, AMBER NetCDF, AMBER mdcrd, TINKER arc and MDTraj HDF5.
MDTraj is based on numpy and is both easy to use and fast. Core routines like RMSD are written in C with explicit SSE vectorization and multicore parallelization. The RMSD code, in particular, is based on the Theobald QCP method and is 4x the speed of the original Theobald code and over 3x as fast as Theobald code modified to use GotoBLAS.
The library also ships with a flexible command-line application for converting trajectories between formats. When you install MDTraj, the script will be installed under the name mdconvert.
Start by loading up a trajectory from disk. MDTraj will automatically parse the file extension and use the appropriate loader.
>>> import mdtraj as md
>>> t = md.load('trajectory.xtc', top='trajectory.pdb')
>>> print t
<mdtraj.Trajectory with 100 frames, 22 atoms at 0x109f0a3d0>
To load files that don’t contain topology information, like Gromacs XTC files, we need to supply something with the top keyword argument that describes the topology, for example a PDB file.
If I’m interest in only a subset of the frames of the trajectory, I can slice it
>>> # lets take a look at the first ten frames
>>> print t[1:10]
<mdtraj.Trajectory with 9 frames, 22 atoms at 0x109b91310>
>>> # or maybe the last frame?
>>> print t[-1]
<mdtraj.Trajectory with 1 frames, 22 atoms at 0x109b97810>
There’s a lot of information in the trajectory object. The most obvious is the cartesian coordinates. They’re stored as a numpy array under xyz. All of the distances in the Trajectory are stored in nanometers. The time unit is picoseconds. Angles are stored in degrees (not radians).
>>> print t.xyz.shape
(100, 22, 3)
>>> print np.mean(t.xyz)
0.89365752249053032
>>> # the simulation time (in picoseconds) of th first 10 frames
>>> print t.time[0:10]
array([ 0.002, 0.004, 0.006, 0.008, 0.01 , 0.012, 0.014, 0.016,
0.018, 0.02 ], dtype=float32)
>>> # or the unitcell lengths in the last frame? (in nanometers of course)
>>> t.unitcell_lengths[-1]
array([ 2., 2., 2.], dtype=float32)
Saving the trajectory back to disk is easy.
>>> # the hdf5 format stores the topology inside the file for convenience
>>> t[::2].save('halftraj.h5')
>>> # the format will be parsed based on the extension, or you can call the
>>> # format-specific save methods
>>> t[0:10].save_dcd('first-ten-frames.dcd')
The trajectory contains a reference to a topology object, which can come in handy. For example, if you want to save a copy of your trajectory with only alpha carbons present, you can do that pretty easily.
>>> atom_to_keep = [a.index for a in t.topology.atoms if a.name == 'CA']
>>> t.restrict_atoms(atoms_to_keep) # this acts inplace on the trajectory
>>> t.save('CA-only.h5')
The source code is licensed under the Lesser GNU General Public License (v2.1+), and available at: https://github.com/rmcgibbo/mdtraj
The best way to report a bug or request a new feature is to make an issue on github. Don’t hesitate to fork the repository, make some changes, and submit a pull request!