Authors: Paul Guénon, Marion Chauveau, Karolina Filipowska, Clément Nizak, Guillaume Stirnemann, Kimberly Reynolds, Olivier Rivoire, Damien Laage
This repository contains the code used for the prediction of the compensatory mutation within the allosteric network of E. coli DHFR.
The method is named COMBINE for COmpensatory Mutations via Boltzmann machine INference of Epistasis. The general idea is to identify positive epistatic interactions, i.e. mutations that are predicted to be significantly more favorable, or less deleterious, in the context of a given mutant than in the wild-type background.
For a detailed description of the COMBINE method, please refer to the following thesis manuscript:
Marion Chauveau. Generative models for protein sequences. PhD thesis manuscript, 2025.
Data/— Contains the MSA and the models used for prediction. Note that the MSA is taken from Thompson et al. 2020 (https://doi.org/10.7554/eLife.53476).utils_COMBINE.py— Includes the main functions used to predict compensatory mutations.COMBINE_DHFR.ipynb— Jupyter notebook that reproduces the analyses and generates the figures shown in the paper.
The models provided in this repository were inferred using the code available in a separate repository: https://github.com/StatBio/Stochastic-Boltzmann-Machine.
Only the resulting models are included here.
If you use this code or data, please cite the associated publication:
Guénon, P., Chauveau, M., Filipowska, K., Nizak, C., Stirnemann, G., Reynolds, K., Rivoire, O., & Laage, D.
Allostery without Large Motions: Molecular Dissection of a Minimal-Shift MWC Allosteric Regulation.