A program designed to select poses produced by molecular dynamics (MD) simulations. This project takes advantage of the MDTraj library to analyze a wide range of outputs from different MD software, and SciKitLearn's Agglomerative Clustering packages for machine learning.
This program is still in development, and we're still working hard to improve it. Right now it works well for simple systems, in particular enzymes with a single active site or with multiple active sites that preform the same function. Work is still being done to evaluate preformance on more complex systems.
Run this like a python script in a terminal window and follow the prompts. You will need:
- A folder containing trajectory files
- Supported file formats include:
pdb,xtc,trr,dcd,binpos,netcdf,mdcrd,prmtop,gsd... and more - Find more information about supported formats here
- Supported file formats include:
- A PDB file which describes the system as it appears in the first frame
- This is for labelling purposes - a raw PDB may not reflect preprocessing you have done
- Innacurate PDB files will not throw an error unless there is a mismatch in the number of atoms
- A list of atom types of interest
- A sample list is included here (based on AMBER / NAMD atom names) but you may supply your own
- The residue name for your subtrate OR residue of interest
- You may get better results from residues from your substrate, depending on the system
If you have feedback or if you encounter an error, please feel free to report it using GitHubs Issues Tab. We check this page regularly and will do our best to attend to errors as quickly as we can.