Protein and RNA structure, function and fitness predicted from evolutionary sequence covariation

Webservers and resources

Evolutionary couplings can be used to predict many interesting aspects about protein and RNA molecules from sequence alone. Here is what we have worked on so far:

EVfold webserver

EVfold server

Compute evolutionary couplings from sequence alignments and predict 3D structure for your protein of interest.


Model database


Precomputed evolutionary couplings and 3D models for thousands of experimentally unsolved proteins.

EVfold models

EVcomplex webserver

EVcomplex server

Predict interacting residues in protein complexes from sequence covariation for your complex of interest.


EVmutation database


Context-dependent mutation landscapes predicted for thousands of human proteins.


Disorder database

Protein plasticity

Coevolutionary prediction of conformations of disordered regions in the human proteome.

Disorder predictions

RNA supplement

RNA structure

3D structure of RNA molecules and protein-RNA interactions predicted from nucleotide sequence covariation.

RNA predictions

Software and source code

We provide access to our evolutionary couplings methods as open source software. For details about each tool, please refer to the individual repositories.


Application and Python package providing the complete EVcouplings workflow from alignment to structure.

Request alpha version


Software to infer undirected pairwise graphical models from multiple sequence alignments.

GitHub repository


Python module to predict mutation effects from undirected graphical models of sequences.

GitHub repository


Tool to interactively visualize the parameters of undirected graphical models of protein families.

GitHub repository