PyLipID - Analysis of Protein–Lipid Interactions from Molecular Dynamics
PyLipID is a Python package for analysing lipid–protein interactions from Molecular Dynamics (MD) simulations. It supports coarse-grained and all-atom simulations, handles formats supported by MDTraj, and works in scripts or Jupyter notebooks.
Documentation: https://pylipid.readthedocs.io
Features
Dual-cutoff contact detection to improve coarse-grained accuracy.
Binding site detection using NetworkX Louvain community analysis.
Kinetic parameters: koff and residence times per residue and site.
Interaction metrics including occupancy, count, duration.
Representative bound pose extraction and clustering.
Binding site surface area (SASA) calculations.
Publication-ready plots and PyMOL scripts.
Installation
From PyPI:
pip install pylipid
From source:
git clone https://github.com/pstansfeld/PyLipID cd PyLipID pip install .
Editable/development install:
pip install -e .
Requires Python >= 3.11 (tested on 3.11–3.14).
Quick Start
from pylipid.api import LipidInteraction
trajfile_list = ["run1/protein_lipids.xtc", "run2/protein_lipids.xtc"]
topfile_list = ["run1/protein_lipids.gro", "run2/protein_lipids.gro"]
li = LipidInteraction(
trajfile_list,
topfile_list=topfile_list,
cutoffs=[0.55, 0.8],
lipid="CHOL",
nprot=1,
)
li.collect_residue_contacts()
li.compute_residue_duration()
li.compute_residue_koff()
li.compute_binding_nodes(threshold=4)
li.compute_site_koff()
li.plot("Residence Time")
li.save_data("Dataset")
li.save_pymol_script(pdb_file="receptor.pdb")
PyLipID runs its calculations in parallel worker processes. When running it as a script, put the code under if __name__ == "__main__": (required on macOS and Windows, where the script would otherwise fail or hang); Jupyter notebooks need no change.
What’s New in v1.6.1
Documentation and packaging only; results are unchanged from v1.6.0. The README and docs cite the published JCTC paper, documentation links point at this repository, and the package lists its current maintainer.
What’s New in v1.6.0
Changes to Results
Binding site surface areas (compute_surface_area) now use the MARTINI bead radii PyLipID has always documented (BB 0.26 nm, SC1-SC4 0.23 nm) and any radii given. Before, mdtraj matched these against element symbols, so they were ignored and the radii of boron and sulfur (0.19/0.18 nm) were used. Each frame is also calculated separately, working around an mdtraj issue that adds part of the previous frame’s area to each frame in a multi-frame call. Site surface areas are typically 5-20% smaller than in earlier versions.
With np.random.seed() set in the calling script, bootstrapped koffs and the poses chosen to represent pose clusters are now reproducible (worker processes previously ignored the seed). Note that with scipy 1.17 (the last release for Python 3.11) curve_fit does not always return the same fit for the same data, so koff values can still vary slightly between runs there; scipy >= 1.18 (Python >= 3.12) is repeatable.
Bug Fixes
analyze_bound_poses no longer fails when many bound poses are identical (DBSCAN eps of 0), and a min_samples passed to it is now used instead of being overwritten.
Corrected binding-site detection off-by-one error.
Fixed calculate_scores returning None.
Fixed sparse_corrcoef matrix type issues.
Fixed DataFrame column/index mismatches in analyze_bound_poses.
Fixed missing return in compute_residue_lipidcount.
Fixed a crash when analyze_bound_poses was run more than once with pandas 3.
cluster_KMeans now always returns labels (it returned a tuple when there were fewer poses than clusters).
plot, plot_logo, save_coordinate and write_site_info calculate missing residue data on demand instead of raising KeyError.
Fixed write_PDB errors for .gro files lacking chain IDs.
Fixed figure overwriting in plot_koff.
Updated deprecated Matplotlib calls.
Fixed LogNorm crash on zero-containing matrices.
Replaced np.float128 with np.longdouble.
Corrected output directory typo.
Performance
collect_residue_contacts computes distances in large batches instead of once per lipid, and shares the batches between worker processes that read the coordinates from shared memory (num_cpus; based on Alex Moger’s parallel contacts in PR #1). Results are identical to the serial calculation; on a 2 x 2,000-frame system contacts take 8 s instead of 40 s.
Bound-pose scoring evaluates the kernel density estimate in numpy (~20x faster; same scores to within floating-point rounding).
Surface areas are shared between worker processes by chunks of frames rather than by trajectory.
Progress bars for parallel steps now follow the calculation.
Survival-function calculation for koff is vectorised (identical results, ~20% faster koff).
Nested progress bars implemented.
Dependencies
Removed python-louvain; using NetworkX built-in Louvain.
Removed statsmodels and p_tqdm (no longer used).
Minimum versions added.
Dropped Python < 3.11; supports Python 3.11–3.14, numpy 2, pandas 3 and current mdtraj.
Packaging & CI
Migrated to pyproject.toml.
CI moved from Travis CI to GitHub Actions.
Citation
@article{song_pylipid_2022,
author = {Song, Wanling and Corey, Robin A. and Ansell, T. Bertie and
Cassidy, C. Keith and Horrell, Michael R. and Duncan, Anna L.
and Stansfeld, Phillip J. and Sansom, Mark S.P.},
title = {PyLipID: A Python Package for Analysis of Protein-Lipid
Interactions from Molecular Dynamics Simulations},
journal = {J. Chem. Theory Comput.},
volume = {18},
number = {2},
pages = {1188--1201},
year = {2022},
doi = {10.1021/acs.jctc.1c00708},
url = {https://doi.org/10.1021/acs.jctc.1c00708},
}
License
MIT License. See LICENSE.txt.
Metadata
Release files for pylipid 1.6.1
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| pylipid-1.6.1.tar.gz | 64.2 kB | Without provenance | Details for pylipid-1.6.1.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| pylipid-1.6.1-py3-none-any.whl | Python 3 | none | any | Details for pylipid-1.6.1-py3-none-any.whl |
Total release size: 137.8 kB
Release files / pylipid-1.6.1.tar.gz
| Download URL | pylipid-1.6.1.tar.gz |
|---|---|
| Size | 64.2 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
01578958657c0f1d4cc5437855db7bae9c19708cc40ad3e99ef93edb9dc71641
|
|
BLAKE2b-256 checksum How to use checksums |
6c622f0d5ff7181f362a925ced73d84b981073131c13eae458ed1c28ad474637
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.12.12
|
Release files / pylipid-1.6.1-py3-none-any.whl
| Download URL | pylipid-1.6.1-py3-none-any.whl |
|---|---|
| Size | 73.7 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
c1cded0e6d64e333a55872cac2cfbb8a76f308bd8087c4250dfe204cab1da3ae
|
|
BLAKE2b-256 checksum How to use checksums |
956a9041428ef45357b90cf263f438e65c628e1e4d1b57a7d8e356dc96295462
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.12.12
|
Security for pylipid 1.6.1
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.