Skip to content

jklsnt/essential-dynamics

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 
 
 

Repository files navigation

protein structural and dynamics analysis

Houjun L., David F., Dhruv C., Ella L.

This package of Python scripts attempts to analyze the dynamics of proteins based on their structure. We hope to show how sequence-level structural changes projects onto the ultimate tertiary-structure molecular dynamics behavior (and therefore eventually the function) of the proteins.

To achieve this, we have to be able to analyze the primary behavior of the protein from simply its sequence. This is done in three steps:

  1. The given re-engineered sequence is re-folded using ESMFold, to yield a PDB structure
  2. The sequence is then analyzed for its general dynamic behavior via the AMBER force field, via openmm, to produce a Fortran DCD trajectory
  3. Finally, the correlations between primary motions per-frame, per-atom in the DCD trajectory is calculated; with its right eigenvalues used for PCA analysis for primary behavior via mdanalysis; the first component (usually accounting for $70\%$ to $90\%$ of the variance) is then projected back onto the backbone of the protein to create a new set of PDB and DCD files

To achieve all of this, we depend on esm, mdanalysis, openmm, openmm-setup, which all can be installed using Conda.

The files in this repo represent the work on each step:

  1. esm/ — folder containing ESM run script and corresponding Dockerfile
  2. run_openmm_simulation.py — molecular dynamics simulations and its initial conditions
  3. pca_analysis.py — PCA analysis of a trajectory, along with projection onto the backbone