Research
We develop and apply computational models, mostly molecular dynamics and statistical learning, to understand how biomolecules move and interact, and what this means for drug action and disease.
The geometry of protein motion
How many degrees of freedom does a protein really use? We estimate the intrinsic dimension of simulated dynamics and design collective variables to describe, compare and bias them, contributing tools to the PLUMED ecosystem.
Key papers: JCIM 2026 · JOSS 2019 · JCP 2025 · Nat Methods 2019
GPCR dynamics and allostery
G protein-coupled receptors are the largest family of drug targets. With the GPCRmd community we run and analyse large collections of receptor simulations to map their conformational dynamics, cryptic allosteric pockets and the lateral gateways through which lipids and ligands enter the transmembrane bundle.
Key papers: Nat Methods 2020 · Nat Commun 2025 · JCIM 2022 · CSBJ 2024
Machine learning and large MD datasets
We build simulation datasets and deep-learning frameworks for data-driven biophysics, from the mdCATH collection of protein-domain trajectories to machine-learned coarse-grained potentials and unsupervised maps of binding pathways.
Key papers: Sci Data 2024 · Nat Commun 2023 · JCTC 2021 · JCP 2024
Binding, misfolding and disease
Together with experimental collaborators we study drug binding and unbinding pathways, amyloidogenic protein variants and the structural impact of mutations, including the dynamics of the SARS-CoV-2 proteome.
Key papers: NAR 2022 · CSBJ 2021 · Protein Sci 2026 · Front Cell Infect Microbiol 2024
Computational statistics
We develop alignment methods for time series, such as the widely used dynamic time warping packages for R and Python, and statistical models for clinical prognosis.
Key papers: JSS 2009 · Leukemia 2017 · Paleoceanogr Paleoclimatol 2025
Images from our papers (see the home page for credits); GPCR figure: Aranda-García et al., Nat Commun 2025, Fig. 2 (CC BY-NC-ND).
See also our software and the full list of publications.