Back

Ensemblify: a user-friendly tool for generating ensembles of intrinsically disordered regions of AlphaFold and user-defined models

Fernandes, N.; Gomes, T. L.; Cordeiro, T. N.

2025-08-30 bioinformatics
10.1101/2025.08.26.672300 bioRxiv
Show abstract

MotivationIntrinsically disordered proteins (IDPs) and regions (IDRs) challenge structural characterization due to their dynamic conformational ensembles and lack of stable structure. Existing computational approaches for modelling these ensembles are either computationally intensive, limited in flexibility, or inaccessible to non-experts, especially when dealing with multi-domain or multi-chain proteins. ResultsWe present Ensemblify, an open-source, user-friendly Python package for generating and analyzing conformational ensembles of IDPs/IDRs. Ensemblify uses a Monte Carlo algorithm coupled with neighbour-aware sampling of dihedral angles from curated or user-defined fragment libraries to explore conformational space. It directly incorporates information from AlphaFolds confidence metrics as flexible energy restraints in PyRosetta to guide the sampling. It supports multi-chain and multi-domain proteins and can sample N-terminal, C-terminal, and inter-domain linkers while preserving folded regions. Ensemble quality can be validated and refined against experimental data such as SAXS via Bayesian/Maximum Entropy (BME) reweighting. Interactive dashboards provide in-depth structural analysis and comparison. Testing across 10 diverse proteins demonstrated Ensemblifys accuracy, flexibility, and ability to recover experimentally observed structural features. Incorporating AlphaFold confidence metrics shows potential to improve the ensemble-data agreement. AvailabilityEnsemblify is freely available at https://github.com/CordeiroLab/ensemblify, along with detailed installation instructions and usage tutorials. Ensemblify can be used for scripting through its Python API or directly through the provided command-line interface (CLI). Complete documentation is available within the source-code and CLI and on Ensemblifys official documentation page (https://ensemblify.readthedocs.io). Contacttiago.gomes@itqb.unl.pt, tiago.cordeiro@itqb.unl.pt Supplementary informationSupplementary data is available online.

Published in Journal of Molecular Biology (predicted rank #5) · training set

Matching journals

The top 2 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.