Shared Haplotype Length Regression and Its Application

The inherent multidimensional nature of complex genetic diseases calls for the development of new statistical methods designed to discover some components of the “missing heritability” problem. Here, we develop and explore the performance of a novel statistical approach based on haplotype sharing concepts. The approach is motivated by the fact that a new variant in a population arising via mutation or migration will reside on a unique haplotype. Assuming the new allele does not become extinct, recombination events at each generation will gradually shorten the haplotype among those with the allele. Thus, on average, relatively recent alleles should reside on longer haplotypes compared to random haplotypes not under positive selection. My statistics employs generalized linear models and generalized estimating equations to test for associations between shared haplotype lengths and traits. Due to the possibility of high number of haplotypes in the model, I also proposed an extension of the statistics that incorporates principal component analysis to alleviate potential collinearity that might inflate the type I error rate of the statistic.

Grantor dc:publisher

Wake Forest University

Year dc:date.issued

2016

Author dc:creator

• Sajuthi, Satria

• Complex Disease

Language dc:language.iso

en

Handle dc:identifier.uri

http://hdl.handle.net/10339/62651

OAI identifier oai:identifier

oai:wakespace.lib.wfu.edu:10339/62651