COMPENSATING CHANGES IN PROTEIN MULTIPLE SEQUENCE ALIGNMENTS

A method was developed to identify compensating changes between residues at positions in a multiple sequence alignment. (For example, one position might always contain a positively charged residue when the other is negatively charged and vice versa.) A correlation-based method was used to measure the compensation found in the four residues at a pair of positions in any two sequences in a multiple alignment. All possible sequence pairings were measured at the pair of positions and the resulting matrix analysed to give a measure of cooperativity among the pairs. The basic method was sufficiently flexible to consider a number of amino acid relatedness models based both on scalar and vectorial properties. Pairs of compensating positions were selected by the method and their mean separation (in a protein of known structure) was compared to both the mean pairwise separation over all residues and the pairwise separation over an equivalent sample of pairs of residues selected on the basis of their conservation alone. The latter is an important control that has been omitted from previous studies. The results indicated that, at best, there was a slight effect (of marginal significance) leading to the selection of closer pairs by the compensation measure when compared to the mean of all pairs. However, this was never as good as the simpler measure based on conservation alone, which always found a significant majority of proteins with a sample mean less than the overall mean.