Modeling Ribosomal Binding Sites and its Application
by
Toby Roseman
Laboratory techniques for determining ribosomal binding sites are tedious and expensive. We model ribosomal binding sites in Escherichia coli using a probability matrix based on three aspects of a binding site: nucleotide sequence of the prospective binding site, start codon, and the distance between the two. Initially the matrix is determined using only the most probable binding sites, then through iterative training the probability matrix is converged to represent a wider range of binding sites.
Observations from recent biochemical experiments have produced the hypothesis that there is selective pressure for motifs which could act as ribosomal binding sites not to appear in the coding regions of genes. Using this model, we provide strong evidence to support this theory.
Advised by Martin Tompa
CSE 403
Wednesday
April 12, 2006
3:30 - 4:20 pm