Email updates

Keep up to date with the latest news and content from BMC Evolutionary Biology and BioMed Central.

Open Access Highly Accessed Methodology article

Detecting groups of coevolving positions in a molecule: a clustering approach

Julien Dutheil* and Nicolas Galtier

Author Affiliations

Institut des Sciences de l'Évolution (UM2-CNRS) Université Montpellier 2 Place Eugéne Bataillon, CC064, 34 095 Montpellier cedex 5, France

For all author emails, please log on.

BMC Evolutionary Biology 2007, 7:242  doi:10.1186/1471-2148-7-242

Published: 30 November 2007

Abstract

Background

Although the patterns of co-substitutions in RNA is now well characterized, detection of coevolving positions in proteins remains a difficult task. It has been recognized that the signal is typically weak, due to the fact that (i) amino-acid are characterized by various biochemical properties, so that distinct amino acids changes are not functionally equivalent, and (ii) a given mutation can be compensated by more than one mutation, at more than one position.

Results

We present a new method based on phylogenetic substitution mapping. The two above-mentioned problems are addressed by (i) the introduction of a weighted mapping, which accounts for the biochemical effects (volume, polarity, charge) of amino-acid changes, (ii) the use of a clustering approach to detect groups of coevolving sites of virtually any size, and (iii) the distinction between biochemical compensation and other coevolutionary mechanisms. We apply this methodology to a previously studied data set of bacterial ribosomal RNA, and to three protein data sets (myoglobin of vertebrates, S-locus Receptor Kinase and Methionine Amino-Peptidase).

Conclusion

We succeed in detecting groups of sites which significantly depart the null hypothesis of independence. Group sizes range from pairs to groups of size ≃ 10, depending on the substitution weights used. The structural and functional relevance of these groups of sites are assessed, and the various evolutionary processes potentially generating correlated substitution patterns are discussed.