The Rockefeller University
Center for Studies in Physics and Biology
2/26/07, 10:00 a.m. - Math Tower Room P 1-122
"Using protein structure to predict nucleosome positions and regulatory
sites in eukaryotic genomes"
Regulation of gene expression is primarily mediated by transcription factors
that bind their cognate DNA sites, and in eukaryotes is strongly affected
by the chromatin structure.
A common approach to inferring transcription factor binding specificities
is to use intergenic sequences from sets of co-regulated genes as input to
the computational methods designed to detect over-represented motifs in
regulatory sequence. The inferred motifs are typically interpreted as the
binding sites, even though there is no direct link
between conserved sequence motifs and the physical chemistry of
protein-DNA interactions. We employed a simple structural model of
protein-DNA binding, the protein structure database, and an extensive
protein localization data set in order to determine factor specificities
in budding yeast with a Bayesian Gibbs sampling algorithm. In comparison
with the best previous study that represents the state of the art in this
area we find that about half of the predicted binding specificities need
revision. We also solve the inverse problem: given a set of co-regulated
genes and a common binding motif, what is the regulatory protein?
Our study reveals a remarkable degree of amino acid conservation at the
protein-DNA binding interface, and points out the utility of enlarging
current structural genomics projects to include all factors with
significantly different DNA binding specificities.
Furthermore, we study the structure of chromatin and its effect on
transcriptional regulation using a DNA mechanics model of the
nucleosome core particle. The model is capable of reproducing in vitro
free energies of nucleosome formation with high accuracy.
We have used this mechanistic approach to predict nucleosome occupancies
in the yeast genome, revealing genome-wide features of nucleosomal
organization. Nucleosome positioning likely affects many chromosomal
functions including transcription factor binding and initiation of gene