Software pipeline for automating omics-scale protein modeling and simulation setup.

• Go to the official online documentation.
• Read a preprint of the paper on bioRxiv.
• See the example dataset from modeling all human tyrosine kinases.

• Daniel L. Parton | [email protected]
• John D. Chodera | [email protected]
• Patrick B. Grinaway | [email protected]

1. Retrieve protein target sequences and template structures.
2. Build models by mapping each target sequence onto every available template structure, using Modeller.
3. Filter out non-unique models (based on a RMSD cutoff).
4. Refine models with implicit solvent molecular dynamics simulation.
5. Refine models with explicit solvent molecular dynamics simulation.
6. (optional) Package and/or compress the final models, ready for transfer or for set-up on other platforms such as Folding@Home.

First go to the Modeller website and get a license key (registration required; free for academic non-profit institutions).

Save the key as an environment variable:

export KEY_MODELLER=XXX

Then, using conda (installs all dependencies except the optional dependency Rosetta):

conda config --add channels omnia
conda config --add channels salilab
conda install ensembler

From source:

git clone https://github.com/choderalab/ensembler.git
cd ensembler
python setup.py install

• OpenMM - https://simtk.org/home/openmm
• Modeller - http://salilab.org/modeller/
• mdtraj - http://mdtraj.org/
• MSMBuilder - http://msmbuilder.org/
• PDBFixer - https://github.com/pandegroup/pdbfixer
• BioPython
• NumPy
• lxml
• PyYAML
• docopt
• mock
• Optional:
  • Rosetta (optional, for template loop reconstruction) - https://www.rosettacommons.org/software
  • MPI4Py (allows many Ensembler functions to be run in parallel using MPI)
  • Pandas (required for certain analysis functions)
  • subprocess32 (if using Python 2)
  • PyMOL (optional, for model alignment/visualization) - http://www.pymol.org/

Recommended approach is to install using conda (https://store.continuum.io/cshop/anaconda/). This will install all dependencies except for the optional dependency Rosetta, which must be installed separately by the user.