Research

To read more about my research and potential research projects for undergraduate students, please see my research statement, attached below.

Linkage Analysis for Categorical Traits

Many complex traits are best defined as categorical traits.  For example, the severity of panic disorder can be quantified based on number of symptoms, and thus has a natural classification as a categorical trait.  While there are well-established methods for locating genes that influence binary or continuous traits, less is known about categorical traits.  I have developed both a Bayesian and a frequentist method for identifying genes for such categorical traits using genetic data from families, and have successfully applied these methods to analyze traits as diverse as panic disorder in humans and Lavender Foal Syndrome in horses.

Ancestry Assignment in Admixed Individuals 

Many people have ancestry from two or more different populations.  For example, many African Americans have some DNA from Africa and some DNA from Europe.  But which DNA is which?  Determining the ancestry of regions of DNA is useful for removing the confounding factor of population stratification in gene-disease association studies, which can otherwise increase type I error.  It's also useful for answering population genetic questions, such as the number of generations since the populations admixed.  In collaboration with Kasia Bryc, I have developed a fast, accurate algorithm that uses Principal Components Analysis (PCA) scores to estimate individuals’ ancestry in windows along each chromosome. 

Association Mapping in Gene Pathways

Traditional methods of gene analysis treat the genes as independent, but in reality, genes influence one another via pathways that can be modeled using graphs.  Taking advantage of the graph structure of a gene pathway has the potential to improve the ability to detect associations between a trait and genes in the pathway.  I have compared five Bayesian prior models for the correlations between the effects of neighboring genes in a pathway using simulated and real data from two pathways from PharmGKB, a database of pathways that affect the body’s response to drugs.

Analysis of Rare Variants

Why do many heritable traits remain incompletely explained by known genetic locations?  Traditional methods of association testing have low power to detect rare genetic variants of small effect.  There are several methods available to counteract this problem by analyzing sets of rare variants within a gene.  However, the majority of these methods assume that the direction of effect is the same for all rare variants, so they are subject to a loss of power when a gene contains variants that both increase and decrease height, for example.  Brooke Fridley and I developed a new method for analysis of rare variants, the Difference in Minor Allele Frequency (D-MAF), which makes no assumptions about the direction of effects.