Novel techniques for statistical and mathematical analyses of sequence data

Research biologists frequently use computer algorithms to compare and match long DNA strings built from the letters C, A, T and G, or proteins built from amino acids. Existing algorithms can fail in certain commonly occurring biological situations where concordance between otherwise closely related sequences is broken at intervals by operations such as shuffling or splicing. This proposal is a continuation of a successful project to develop new mathematical methods for these situations, built on the statistical properties of short word frequencies within the sequences. Similar principles will also be applied to developing algorithms for locating regions within the genome responsible for regulating the expression of genes.