Services are gradually being migrated to
Please try out the new site.


InterMap3D predicts interacting protein residues by identifying co-evolving pairs of aminoacids from an alignment of protein sequences. The only required input is a single protein sequence in fasta format, but using a manually curated alignment is advised. If you nevertheless only input a single protein sequence, then an alignment will be generated automatically and, unless you choose another option, realigned with MAFFT. We advise caution in interpreting results from an automatically generated alignment, and that the user examines the alignment itself.
Once an alignment is available (either provided by the user or automatically generated), it is used for detecting coevolving positions, which can be done by one or more of the following three methods: "RCW MI", "MI/Entropy" and "Dependency". Moreover, by selecting more than one method (checking in more than one checkbox) one can obtain the intersection of the predictions of 2 or 3 methods. This can be helpful in obtaining more robust predictions. To obtain each method's predictions separately, one should run the server several times, one for each method.
These predicted positions are then depicted on the 3D structure of the protein most similar to yours available in the PDB database.

If you mouse-over the options, you will get a short explanation box. For a more detailed explanation, please refer to the instructions page.

Instructions Output format Article abstract


Paste a single sequence or an alignment of several sequences in FASTA format into the field below:
(for better results, use a manually curated alignment with plenty of variation and more than 50 sequences)

Submit a file in FASTA format directly from your local disk:


Options on alignment generation Consider % the minimum length of the query protein that has to be matched when looking for homologues to build an alignment

Alignment Methods:

Options on coevolution analysis Consider the   highest scoring hits from InterMap3D in the output.

P-value Calculation
Number of simulated alignments for calculating a null score distribution:

For RCW MI only: Consider up to   residues in the coevolving pairs (2), triplets (3), etc. Minimum value is 1.

Options on visualization of results Colour scheme:
No graphics
Black background
White background

Residue numbering:
Number by the query protein
Number by the alignment

The visualization of results is provided by FeatureMap3D, where more details can be found.

Hit Submit to run the search with default parameters, or customize the settings above
Expected running time: appx. 5 minutes per sequence.


For publication of results, please cite:

InterMap3D: visualizing predicted co-evolving protein residues.
R. Gouveia-Oliveira, F. Roque, R. Wernersson, T. Sicheritz-Ponten, P. W. Sackett, A. Mølgaard, A. G. Pedersen.
Bioinformatics,Vol. 25, no. 15, 2009, pages 1963-1965. Download full article.

As well as any of the ones below you may have used:

Finding coevolving amino acid residues using row and column weighting of mutual information and multi-dimensional amino acid representation
R. Gouveia-Oliveira, A. G. Pedersen.
Algorithms in Molecular Biology 2007 Oct 3;2(1):12 .

Using information theory to search for co-evolving residues in proteins.
L. C. Martin, G. B. Gloor, S. D. Dunn and L. M. Wahl
Bioinformatics,Vol. 21, no. 22, 2005, pages 4116-4124

Using multiple interdependency to separate functional from phylogenetic correlations in protein alignments.
Elisabeth R. M. Tillier and Thomas W. H. Lui
Bioinformatics,Vol. 19, 2003, pages 750-755

As new methods are added, additional references to be cited will be placed here.


InterMap3D allows the introduction of new methods for predicting co-evolving residues. If you have developed a method in this area and would like to see it integrated with InterMap3D, please write us by email to


Scientific problems:        Technical problems: