Creative Commons source reading · Prediction and forecasting
AlloPool is a deep learning framework that infers protein allostery from molecular dynamics simulations
Marfoglia M; Pedraza-Joya MA; Guirardel L; Chatzi Souleiman A; Barth P
Originally published in PLoS Biol · 2026-09-21
From the article
Selected source text. JournalOS removes markup and shortens excerpts where indicated. The full article provides the complete context.
Abstract
AbstractRecent advances in artificial intelligence have transformed protein structure prediction and design. However, protein function is governed not only by static structures but also by the conformational dynamics that allow proteins to access distinct functional states. Predicting these dynamic transitions, central to many biological processes, remains challenging due to the scarcity of high-resolution experimental data, which limits the training of machine-learning models for dynamic and energetic property prediction. Here, we present AlloPool, a graph neural network (GNN)-based framework that interprets molecular dynamics simulations by iteratively pruning residue–residue interactions to uncover minimal, time-resolved interaction networks that govern protein conformational dynamics and structural responses to chemical or…
Discussion
Protein motions are fundamental to function, yet they remain difficult to study, model and analyze. To address this challenge, we developed and validated AlloPool, a deep learning framework that establishes a new standard for interpreting protein dynamics from molecular simulation data. Unlike existing approaches, AlloPool is explicitly designed to learn time-dependent conformational changes and the sparse interaction networks that drive structural transitions. As a result, AlloPool accurately predicts complex dynamic behaviors, including time-resolved fluctuations in allosteric communication, protein binding and folding processes from molecular dynamics trajectories. AlloPool was rigorously validated across a diverse range of dynamic protein systems, including small soluble domains, membrane receptors, mechanosensors, and…
Editorial reading guide
- What data created and tested the model, and was evaluation independent?
- What is the cost of a false positive, false negative or out-of-distribution decision?
