CQSB · Sorbonne Université · CNRS

Statistical Genomics and Biological Physics

We work at the intersection of statistical physics, computational biology and machine learning. Much of our research starts from families of protein or RNA sequences and asks what they can tell us about structure, function and evolution.

Starting from coevolution and inverse statistical physics, we develop generative models of biological sequences and use them to study epistasis, evolution and sequence design. We also combine models with targeted experiments, and more recently work on inference problems in DNA microscopy and spatial molecular biology.

Graphical abstract showing natural sequences used to train a generative model, sampled sequences tested experimentally, and experimental results reintegrated into a new model.
Generative modelling with experimental feedback. Graphical abstract from Calvanese et al., Nucleic Acids Research (2025).

Research

Sequence statistics

Coevolution and inverse statistical physics

We infer direct statistical couplings from protein and RNA families and use them to study contacts, molecular interactions and mutational constraints.

Generative models

From DCA to machine learning

Potts models, autoregressive models and other generative approaches let us describe whole sequence families rather than individual sites or mutations.

Evolution and design

Epistasis, evolvability and new sequences

We study how proteins move through sequence space, how robustness and specificity change along evolutionary paths, and how evolutionary information can guide sequence design.

Spatial inference

DNA microscopy and molecular graphs

Here the raw data are not images but noisy proximity networks. We develop statistical and machine-learning methods to infer the underlying spatial organization.

People

Current team

  • Martin WeigtProfessor · group leader
  • Pierre Barrat-CharlaixMaître de conférences
  • Francesco CalvanesePostdoc
  • Leonardo Di BariPhD student
  • Fabio GiacomettiPhD student
  • Roberto NettiPhD student
  • Giovanni PeinettiPhD student
  • Elena Rodriguez HortaPhD student
  • Lorenzo RossetPhD student
  • Federico ScarpatiPhD student
  • Leonardo Trevisan MotaPhD student
  • Alya ZeinatyPhD student

Former members

  • Matteo Bisardi
  • Sabrina Cotogno
  • Alice Coucke
  • Giancarlo Croce
  • Christoph Feinauer
  • Matteo Figliuzzi
  • Carlos Gandarilla
  • Eleonora de Leonardis
  • Barthélémy Meynard
  • Anna-Paola Muntoni
  • Maureen Muscat
  • Edwin Rodriguez Horta
  • Juan Rodriguez-Rivas
  • Edoardo Sarti
  • Kai Shimagaki
  • Jeanne Trinquier
  • Guido Uguzzoni

Selected papers

2026 · Machine Learning: Science and Technology
Steering Sequence Generation in Protein Language Models through Iterative Lookback Monte Carlo Sampling
Calvanese, Lombardi, Weigt & Fernandez-de-Cossio-Diaz
2025 · Nucleic Acids Research
Integrating experimental feedback improves generative models for biological sequences
Calvanese, Peinetti, Pavlinova, Nghe & Weigt
2024 · Nucleic Acids Research
Towards parsimonious generative modeling of RNA families
Calvanese, Lambert, Nghe, Zamponi & Weigt
2024 · PNAS
Emergent time scales of epistasis in protein evolution
Di Bari, Bisardi, Cotogno, Weigt & Zamponi
2022 · Nature Communications
Deciphering polymorphism in 61,157 Escherichia coli genomes via epistatic sequence landscapes
Vigué, Croce, Petitjean, Ruppé, Tenaillon & Weigt
2020 · Science
An evolution-based model for designing chorismate mutase enzymes
Russ, Figliuzzi, Stocker et al.

Contact

Statistical Genomics and Biological Physics
Computational, Quantitative and Synthetic Biology (CQSB), UMR 7238
Sorbonne Université · CNRS
4 place Jussieu
Building C, room C414
75005 Paris, France