Sorbonne Université · CQSB

Martin Weigt

Professor of Computer Science
Statistical physics · Computational biology · Machine learning

Martin Weigt

I am a statistical physicist by training, and much of my work asks what can be learned from biological sequences. I came to this through coevolution: correlated mutations in proteins can reveal contacts within structures and interactions between molecules. This led us to Direct Coupling Analysis and, later, to broader generative models for proteins and RNA.

Today I use these models to study how sequence relates to structure and function, how epistasis shapes molecular evolution, and how evolutionary information can help us design new sequences. I am especially interested in situations where experimental data are scarce: rather than relying only on larger models, we try to combine sequence statistics with biophysical constraints and targeted experiments.

A newer direction is spatial molecular biology. In DNA microscopy, for example, the data are not images but noisy molecular proximity graphs, and the challenge is to infer spatial organization from them.

Research

01

Coevolution and generative models

We use correlations in protein and RNA families to learn about contacts, interactions and mutational effects. This includes DCA and Potts models, autoregressive models and other generative approaches.

02

Evolution, epistasis and design

I am interested in how proteins move through sequence space: how epistasis, robustness and specificity shape evolution, and how the same ideas can be used to generate and test new sequences.

03

Spatial molecular biology

In DNA microscopy and related approaches, we infer spatial structure from sequencing-based proximity data. The problems combine statistical inference, graph methods and machine learning.

From coevolution to generative models

Some selected papers.

Contact

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