Investigate water in MD trajectories.
Install using python setup.py install
This package was written and tested on Python 3.4. Python 2 is supported through the python-future package. Let me know if you run into any issues.
Setup will install new subcommands to the msmb
command.
-
SolventShellsFeaturizer
: Featurize trajectories. Use it like any other MSMBuilder featurizer. -
SolventShellsAssigner
: Used for generating "assignments" of atoms to shells for visualization -
SolventApplyComponents
: We can leverage the linear coefficients of decomposition models like PCA or tICA to enhance visualization of "important" solvent atoms. This command takes as input the result ofSolventShellsAssigner
(which assigns solvent atoms to features) as well as a serialized decomposition model (PCA or tICA) and weights solvent atoms by the coefficients of the features to which they are assigned. -
SolventWriteVMD
: This will export data from the above command in a format readable by VMD. This command will also generate a script which loads the data into theuser
field in VMD. This allows, e.g., coloring and selection based on solvent tICA coefficients.
"Solvent" atoms are binned into shells surrounding "solute" atoms. The count of atoms in each shells can be used as a feature vector for use with MSMBuilder. By assigning solvent atoms to specific features, by identifying important features we can actually identify individual, relevant solvent atoms. For example, in the figure above, the bright-red "solvent" atom is contained within two shells (features) identified as important (perhaps by tICA). Using the provided scripts, one can easily perform such a visualization in VMD.