Relating genotype to phenotype: Resampling-based multiple hypothesis testing using order statistics.

Jennifer Schumi
Biostatistics
Harvard University
*Candidate for the Assistant or Associate Professor position

Friday, February 10th
10:00am
Moos 2-690
Minneapolis Campus

Abstract:

Anti-retroviral drugs have had a profound effect on the management of the HIV epidemic in many countries, but genetic mutations in the virus can lead to resistance and limit the effectiveness of these treatments. Physicians and clinical researchers are interested in statistical methods that relate genotype patterns to a measure of drug sensitivity phenotype in order to assess the impact of the resistance mutations. We have developed a semi-parametric resampling-based approach to multiple hypothesis testing that identifies patterns of mutations associated with changes in drug susceptibility with respect to the wild-type virus. The use of order statistics in our method is similar in spirit to normal probability plots, but does not require large sample sizes in each genotype pattern to ensure approximate normality. Additionally, our approach has been shown to be more powerful and more efficient at classification than traditional multiple testing methods such as the Bonferroni or Benjamini-Hochberg corrections. Two applications will be presented; the first identifies patterns of mutations that enhance or decrease drug susceptibility to the protease inhibitor Amprenavir; the second investigates interactions between mutations. These analyses allow the investigation of how genetic mutations act in the presence of other mutations and may suggest mechanisms by which anti-retroviral resistance occurs or is reversed through the accumulation of genetic mutations.
A social tea will be held at 9:30A.M. in A434 Mayo. All are Welcome.
For more details contact 612-624-4655 or see http://www.biostat.umn.edu/seminar_academic.html