Motivation: The constituent amino acids of a protein work together to define its structure and to facilitate its function. Their interdependence should be apparent in the evolutionary record of each protein family: positions in the sequence of a protein family that are intimately associated in space or in function should co-vary in evolution. A recent approach by Ranganathan and colleagues proposes to look at subsets of a protein family, selected for their sequence at one position, to see how this affects variation at other positions.

Results: We present a quantitative algorithm for assessing covariation with this approach, based on explicit likelihood calculations. By applying our algorithm to 138 Pfam families with at least one member of known structure, we demonstrate that our method has improved power in finding physically close residues in crystal structures, compared to that of Ranganathan and colleagues.

Supplementary information:www.afodor.net/bioinfosup.html

To whom correspondence should be addressed.
The authors wish it to be known that, in their opinion, these two authors should be regarded as joint First Authors.

Author notes

1Department of Neurobiology, Harvard Medical School, 220 Longwood Avenue, Boston, MA 02115, USA and 2Department of Molecular and Cellular Physiology, Howard Hughes Medical Institute, Stanford University School of Medicine, Stanford, CA 94305-5345, USA