Trajectory¶
Try pytraj online:
See also¶
- Fancy indexing of trajectory
- Trajectory warm up
- Register to load from disk
- Register to load from disk with frame stride
- Register to load several files from disk
- Register to load several files from disk with frame stride
- Write to disk
- Extract coordinates from PDB structure
- Write PDB file to netcdf format (require Netcdf lib)
- Fancy indexing
- Iterate whole trajectory
- Iterate a part of trajectory
- Perform calculation
- Convert data to pandas DataFrame
- Convert to different file format
- Combine with cpptraj command style
- Convert out-of-core TrajectoryIterator to in-memory Trajectory
- Parallel computing with TrajectoryIterator
- Modify trajectory
Overview¶
There are two types of trajecoties in pytraj.
- an immutable
pytraj.TrajectoryIterator(work-horse of pytraj).
This class offers out-of-core data store with easiness to load data to memory.
- a mutable
pytraj.Trajectory. This class hold in-memory data. It is suitable for loading a chunk of snapshots.
Why are there two trajectory classes?¶
See our rationale here Trajectory design
Try it¶
In [1]: import pytraj as pt
In [2]: traj = pt.iterload('data/tz2.ortho.nc', 'data/tz2.ortho.parm7')
In [3]: traj
Out[3]:
pytraj.TrajectoryIterator, 10 frames:
Size: 0.001183 (GB)
<Topology: 5293 atoms, 1704 residues, 1692 mols, PBC with box type = ortho>
In [4]: pt.rmsd(traj, ref=0, mask='@CA')
Out[4]:
array([ 0. , 0.2641, 0.386 , 0.2688, 0.2619, 0.4576, 0.4565,
0.4443, 0.4363, 0.4475])
pytraj is able to detect single file (mol2, pdb) to load as TrajectoryIterator.
>>> pt.iterload('my_pdb.pdb')
>>> pt.iterload('your_mol2.mol2')
TrajectoryIterator is something like range(start, stop[, step]) in python3 or xrange(start, stop[, step]) in python2
In [5]: for i in range(0, 8, 2): print(i)
0
2
4
6
In [6]: for f in traj(0, 8, 2): print(f)
<Frame with 5293 atoms>
<Frame with 5293 atoms>
<Frame with 5293 atoms>
<Frame with 5293 atoms>
However, TrajectoryIterator is much more than that, you can slice atoms:
In [7]: for f in traj(0, 8, 2, mask='@CA'): print(f, f.xyz[0])
<Frame with 12 atoms> [ 14.8788 27.639 18.088 ]
<Frame with 12 atoms> [ 14.4703 27.9171 18.2456]
<Frame with 12 atoms> [ 14.5707 27.8402 17.4244]
<Frame with 12 atoms> [ 14.7721 28.067 16.9504]
To load all frames to memory at once, use [] notation:
In [8]: traj[0:8:2, '@CA']
Out[8]:
pytraj.Trajectory, 4 frames:
Size: 0.000001 (GB)
<Topology: 12 atoms, 12 residues, 12 mols, PBC with box type = ortho>
How to perform analysis with TrajectoryIterator? It’s very simple. For example, if you want to calculate rmsd to 3rd frame (index starts from 0) for all atoms, just:
In [9]: pt.rmsd(traj, ref=3)
Out[9]:
array([ 11.2707, 10.0133, 8.3305, 0. , 8.288 , 10.0907,
11.2475, 11.9937, 12.0499, 12.3873])
You have TB of data and want to speed up your calculation, just add more cpus
In [10]: pt.pmap(pt.rmsd, traj, ref=3, n_cores=4)
Out[10]:
OrderedDict([('RMSD_00001',
array([ 11.2707, 10.0133, 8.3305, 0. , 8.288 , 10.0907,
11.2475, 11.9937, 12.0499, 12.3873]))])
How to get raw coordinates?
>>> traj.xyz
>>> traj[[1, 3, 5]].xyz