|A Statistical Strategy for Predicting Drug Response with Unphased Genetic Data
Rongling Wu, PhD, Department of Statistics, Pennsylvania State University College of Medicine.
Monday, January 26, 2009
4:00 pm 223 Weber
Increasingly available genetic data hold a great promise to predict drug response for individual patients based on their genetic makeup. In this talk, I will present a statistical strategy for computing genes and genomes for patients and predicting their predisposition to drugs. The central idea of this strategy is to integrate mathematical and biochemical aspects of drug reactions into a mixture framework for haplotype discovery and modeling with unphased genetic data. The effects of haplotypes constructed by DNA sequences on drug response can be quantified by estimating and testing a few parameters that define the shape and pattern of pharmacokinetic and pharmacodynamic processes. The new strategy was applied to analyze a real pharmacogenomic data set, leading to the detection of significant haplotypes that affect heart rate increase in response to dobutamine. Extensive simulation studies were performed to investigate the statistical properties of the design and validate its usefulness and utilization. The new strategy can be used to shed light on the genetic architecture of inter-patient variation in complex drug responses and provide scientific guidance about the design of personalized medicine.