Installable VMD as a python module
NEW: Version 3.0 has the following new features
- Atomselection attributes can be accessed or set more easily:
atomsel.get("x")
can be written asatomsel.x
! - Removed extra info dumped to stdout
- Docstrings for all methods and modules
- Much more functionality as Python methods!
atomsel.hbonds
, etc - The
selection
module lets you define custom atomselection macros - More rigorous reference counting / fewer memory leaks
- Much prettier code
NEW: Support for Python 3!!!
All features of VMD from the 1.9.4 tree, plus some optional plugins not included in binary distributions:
- Read and write formal charges to MAE files
- DMS plugin for DESRES molecular file format
- HOOMD plugin
- It doesn't crash when you import it
- Doesn't care which numpy you compile against
- Support for Python 2 or Python 3
The following sub-modules are part of VMD. The import system
makes more sense now, so standard importing like from vmd import atomsel
will work correctly.
Some of these modules don't make much sense without the graphical display window. I don't currently distribute a VMD binary with Python built in, but you can look at my build scripts and compile your own, or use the available ones from the official developers.
The following python modules are part of vmd-python. Some of these are currently not documented on the website (indicated with *).
- animate
- axes
- atomsel (atom selection language)
- color
- display
- evaltcl (run all your old tcl scripts)
- graphics
- imd
- label
- material
- measure*
- molecule (read and write all kinds of molecules)
- molrep
- mouse* (of limited utility in the command line...)
- render
- trans
- topology*
- vmdmenu*
- vmdnumpy (very very useful!)
There are also some object oriented classes that provide a higher-level interface to the sub-modules above. Note these are case sensitive!!
- Label
- Material
- Molecule
Here's a very simple example of using the VMD-python API to calculate the root-mean-square-
fluctuation of all tyrosine residues in a loaded trajectory molid
relative to some
average structure molid_avg
:
from vmd import molecule, vmdnumpy
import numpy as np
mask = vmdnumpy.atomselect(molid_avg, 0, "resname TYR")
ref = np.compress(mask, vmdnumpy.timestep(molid_avg, 0), axis=0)
rmsf = np.zeros(len(ref))
for frame in range(molecule.numframes(molid)):
frame = np.compress(mask, vmdnumpy.timestep(molid, f), axis=0)
rmsf += np.sqrt(np.sum((frame-ref)**2, axis=1))
rmsf /= float(molecule.numframes(molid)-minframe)
rmsf = np.sqrt(rmsf)
Please refer to the documentation for description of all functions. Also note that the built in help() command in Python may sometimes give you more up-to-date information than the website.
Don't confuse the Molecule and the molecule modules. Molecule is part of the high-level Python interface, and can be pickled and have different representations. It's a class. However molecule is a module that allows loading and saving of molecules while tracking them by an integer index.
The easiest way to install the module is by using the Conda Python package manager. Vmd-python is in the conda forge channel:
Simple binary installation with conda (currently linux-x86_64 only)
conda install -c conda-forge vmd-python
For other architectures, you can build from source by cloning this repo:
python setup.py build
python setup.py install
Installation can take a while since it compiles VMD from source.
NOTE: The build process for this is currently broken.
VMD-Python can render to an offscreen buffer if you build it yourself with the --egl flag. You'll need the following files on your system to build and link successfully:
* GL/gl.h
* EGL/egl.h, EGL/eglext.h
* libOpenGL.so
In Debian, these are provided in libegl1-mesa-dev
, mesa-common-dev
,
and libopengl0
packages. To build:
python setup.py --egl build
python setup.py install
Please file issues if you have problems with this!!
Users have reported that on OSX with Tcl/Tk 8.6 the module will segfault on import. Unfortunately I'm the only developer of this software and don't have access to a machine running OSX with which to debug the issue.
vmd-python has the following dependencies:
* python 2.7 or 3.6
* numpy
* libnetcdf >= 4.3
* expat
* sqlite
* Tcl/Tk = 8.5 (8.6 will crash on OSX)
To build on Debian, that's the following packages:
* libsqlite3-dev
* libexpat1-dev
* libnetcdf-dev
* tcl8.5-dev
* libegl1-dev
* libopengl0
This software is classified as a "derivative work" of the original Visual Molecular Dynamics software. It uses the main VMD source code and Python compilation options as well as code I have written for simple compilation and installation, as well as several patches I have applied.
As such, this is a MODIFIED VERSION of VMD and is not the original software distributed by Illinois.
"This software includes code developed by the Theoretical and Computational Biophysics Group at the Beckman Institute for Advanced Science and Technology at the University of Illinois at Urbana-Champaign."
Official VMD web page: http://www.ks.uiuc.edu/Research/vmd