Some Physical Problems in Bioinformatics

Bioinformatics is a data driven field, in which a significant number of problems require statistical modeling. The flood of data emerging from genome centers uses sequence comparison to delimit and assign function to genes, and in very limited ways infers gene control from approximately repeated sequence motifs near to the genes themselves. Traditional topics in computer science such as coding theory, natural language processing, and old fashioned cryptography all impinge on the problem of deducing regulatory information from the genome, but are not probabalistic enough to cope with the fuzziness of biological patterns. The means by which living things encode information is a problem common to both neural biology and the regulation of gene expression by the genome. Physical analogies are employed to highlight some of the problems and opportunities in this area.