Ensemblify: A Python package for generating ensembles of intrinsically disordered regions of AlphaFold or user defined models
Ensemblify is a Python package that can generate protein conformational ensembles by sampling dihedral angle values from a three-residue fragment database and inserting them into flexible regions of a protein of interest (e.g. intrinsically disordered regions (IDRs)).
It supports both user-defined models and AlphaFold[1] predictions, using the predicted Local Distance Difference Test (pLDDT) and Predicted Aligned Error (PAE) confidence metrics to guide conformational sampling. Designed to enhance the study of IDRs, it allows flexible customization of sampling parameters and works with single or multi-chain proteins, offering a powerful tool for protein structure research. Ensemble analysis and reweighting with experimental data is also available through interactive graphical dashboards.
Step-by-step instructions for installing Ensemblify are available in the Documentation.
Ensemblify can be used either as a Command Line Interface (CLI):
conda activate ensemblify_env
(ensemblify_env) $ ensemblify [options]
or as a Python library inside a script or Jupyter notebook:
import ensemblify as ey
ey.show_config()
Check the Documentation for more details.
A general overview of Ensemblify, descriptions of employed methods and applications can be found in the Ensemblify pre-print and accompanying support information.
Ensemblify provides a three-residue fragment (tripeptide) database from which to sample dihedral angle values.
This database is provided separately from the Ensemblify source-code.
You can get it here and more about its creation in the Documentation.
Ensemblify's documentation is available together with an API reference at https://ensemblify.readthedocs.io. Alternatively, the source-code contains docstrings with relevant information.
If you use Ensemblify, please cite its original paper:
@article {ensemblify2025,
title = {Ensemblify: a user-friendly tool for generating ensembles of intrinsically disordered regions of AlphaFold and user-defined models},
author = {Fernandes, Nuno and Gomes, Tiago Lopes and Cordeiro, Tiago N},
journal = {bioRxiv}
year = {2025},
publisher = {Cold Spring Harbor Laboratory},
doi = {10.1101/2025.08.26.672300},
URL = {https://www.biorxiv.org/content/early/2025/08/30/2025.08.26.672300},
}
We would like to thank the DeepMind team for developing AlphaFold.
We would also like to thank the team at the Juan Cortés lab in the LAAS-CNRS institute for creating the tripeptide database used in the development of this tool. Check out their work at https://moma.laas.fr/.
Nuno P. Fernandes (Main Developer) [GitHub]
Tiago Lopes Gomes (Initial prototyping, Supervisor) [GitHub]
Tiago N. Cordeiro (Supervisor) [GitHub]