|
Abstract:
|
DNA copy number variation (CNV) has received increasing attention over the past several years, and multiple studies have revealed both common inherited variants and unique de novo variants in healthy, as well as diseased, subjects. CNVs have been associated with several disease phenotypes, as well as with expression quantitative trait loci (eQTLs). Most existing methodologies for inferring copy number from microarray data have focused on analyzing arrays one at a time or, if analyzing multiple arrays, on a post-hoc basis. A multilevel hidden Markov model (mHMM) of CNV, however, simultaneously accounts both for the correlation of CNV state at different loci within an individual (i.e. along chromosomes) as well as between individuals. I will introduce a general inferential framework, examine computational and inferential challenges, and present results on simulated and actual data.
|
