Sequence composition approaches

Sequence database clustering requires particularly speedy pairwise comparisons. A number of groups have approached this problem by comparing amino acid composition and/or the distributions of pairs (or triplets etc.) of amino acids or nucleotides[Van-Heel, 1991,Wu et al., 1992, for example]. Recently, Hobohm and Sanderhobohm:propsearch extended this approach to include global characteristics such as sequence length and calculated isoelectric point in addition to amino acid and pair composition. Using an optimised set of weightings for the various measurements, rather inconclusive sequence database searches were performed using multiple sequences as queries. Non-linear mappings of sequence composition data have also been used to cluster large sets of sequences[Ferran et al., 1994,Hanke et al., 1996].