Skip to content

elkins/synth-dynamics

Repository files navigation

synth-dynamics: Time-Resolved Ensemble Generator

codecov PyPI version Python License: MIT Tests Lint Ruff Checked with mypy

synth-dynamics is a fast, lightweight molecular dynamics engine designed to generate meaningful conformational ensembles of proteins. Unlike full-atom simulations (like GROMACS or Amber), synth-dynamics uses a Coarse-Grained Anisotropic Network Model (ANM) and Langevin dynamics to capture the essential global motions of proteins with minimal computational overhead.

This tool is designed to bridge the gap between static structures and time-averaged experimental observables, such as NMR relaxation parameters, SAXS Kratky plots, or FRET efficiency distributions.


🧪 For Structural Biologists

  • Ensemble Generation: Rapidly sample the conformational landscape of a protein to simulate disordered states or flexible loops.
  • Integration: Works seamlessly with MDAnalysis and biotite for trajectory processing.

🤖 For Machine Learning Researchers

  • Dynamic Training Data: Generate synthetic molecular trajectories to train time-series models (LSTMs, Transformers) or dynamic GNNs.
  • High Performance: Optimized for throughput, allowing the generation of millisecond-scale ensembles in seconds.

Key Features

  • Coarse-Grained Simulation: Models proteins using C-alpha atoms and harmonic "spring" networks.
  • Fast Langevin Engine: Propagates coordinates using a stable, overdamped Langevin integrator.
  • Experimental Integration: Perfect for generating the structural ensembles needed for synth-nmr or synth-saxs.
  • Easy to Use: Simple API for loading PDBs, configuring forcefields, and running simulations.
  • Extensively Tested: 100% test coverage ensuring reliability and correctness.

Installation

synth-dynamics requires Python 3.10+ and the following dependencies:

pip install numpy MDAnalysis scipy

To install the documentation theme:

pip install sphinx_rtd_theme

Quick Start

Running a simulation is straightforward:

from synth_dynamics import System, ANMForceField, LangevinIntegrator, Simulation

# 1. Load the system (automatically filters for C-alpha atoms)
system = System("protein.pdb")

# 2. Define the Anisotropic Network Model forcefield
# Cutoff (15A) and spring constant determine the flexibility
ff = ANMForceField(system.equilibrium_coords, cutoff=15.0, spring_constant=1.0)

# 3. Initialize the Langevin integrator (dt in ps, T in Kelvin)
integrator = LangevinIntegrator(dt=0.1, temperature=300.0, friction=1.0)

# 4. Run and save the trajectory
sim = Simulation(system, ff, integrator)
sim.run(n_steps=1000, output_path="trajectory.dcd", stride=10)

Documentation

Full API documentation and usage guides are available in the docs/ directory. You can build the HTML documentation locally:

cd docs
sphinx-build -b html . _build/html

Testing

To run the test suite and verify coverage:

PYTHONPATH=. pytest --cov=synth_dynamics tests/

License

This project is licensed under the MIT License - see the LICENSE file for details (if applicable).

Related Projects

This library is part of the synth-pdb ecosystem — use it to generate trajectories for synth-nmr or synth-saxs ensemble averaging:

Contributing

Contributions are welcome! Please open an issue or pull request on GitHub. Run pre-commit run --all-files before submitting.

Citation

@software{synth_dynamics,
  author  = {Elkins, George},
  title   = {synth-dynamics: Coarse-grained protein dynamics for ensemble generation},
  year    = {2026},
  url     = {https://github.com/elkins/synth-dynamics},
  version = {0.1.1}
}

About

Fast Langevin dynamics engine using the Anisotropic Network Model (ANM) for rapid protein conformational ensemble generation.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors