Martinize.py is a python script to convert (atomistic) protein structures to Martini coarse grain (CG) structures and topologies.
It has been originally been written by Djurre H. de Jong and Tsjerk A. Wassenaar and is currently maintained Djurre H. de Jong and Jaakko. J. Uusitalo. The script is written to be a useful tool, but comes with no warranty. If you use the script for any of your publications, please cite:
de Jong et al., J. Chem. Theory Comput., 2013, DOI:10.1021/ct300646g
martinize.py is provided as a single-file script and no installation is needed other than downloading the script. The newest, tested version of the script can always be downloaded from:
Possible input arguments can be learned by typing "martinize.py -h" on the commmand-line. Notice that the script needs a working version of python 2. A python 3 version is available upon request.
Although the final script comes as a single file, internally it consists of several "Modules". In the development version the modules all come in seperate files. The one-file-script can be created by running "catenate.py".
Organization of the different modules:
Force field parameters are specified in the differentt forcefield modules, e.g.: martini22_ff.py. In the development version all forcefields are placed in the "ForceFields/" subdirectory.
When catenating the modules to the final script, one gets to choose which forcefields to include. When later running the script one can use these selected forcefields AND additional forcefields present in the working directory. The latter allows for flexible implementation of own forcefields.
Each version of martinize.py is tested for a number of protein structures, trying to cover as many aspect of its function as possible. The testset is compiled in single script in the "tests/" subdirectory. It should be checked that all stuctures run without error and the obtained structure and topology files do not differ from previous versions.
It is impossible to test every possible usage scenario, and currently especially the multiscale functionalities are lacking.