Statistical Methods for Integrating Multiple Sources of Genomic Data

Erin Conlon
University of Massachusetts

Wednesday, March 28, 2007
3:30pm
Moos 2-690
Minneapolis Campus

Abstract:

Erin Conlon will present two statistical approaches for combining multiple sources of genomic data. First, she will describe a Bayesian meta-analysis model that combines multiple cDNA micro-array studies to identify over-expressed genes. This model produces the gene-specific posterior probability of differential expression as well as Bayesian estimates of False Discovery Rates. We find in simulations that this model improves results verses individual study analyses.

Erin will then describe a novel statistical and bio-informatic method, Motif Regressor, that identifies motifs in regulatory DNA sequence using the combination of micro-array and DNA sequence data. Motifs are statistically tested for association between motif occurrences and global gene expression patterns in order to identify significant regulatory motifs. An Additive linear model determines motifs acting concurrently to control gene expression. Model organisms yeast and bacteria are used to illustrate these methods.

A social tea will be held at 3:00 P.M. in A434 Mayo. All are Welcome.
For more details contact 612-624-4655 or see http://www.biostat.umn.edu/seminar_academic.html