|
Abstract:
|
In many genetic epidemiology studies, clinical assessments are recorded using either categorical or more precisely ordinal data. For example, doctors may record intellectual or motor disability as normal, mild, moderate, or severe. Moreover, critical information may be lost if we reduce our data to a simply binary trait, e.g., normal and not normal. over the past decade, group tests have been widely developed to detect the association of rare genetic variants in sequencing studies. However, most statistical procedures are designed for either continuous or binary outcomes. In this article, we present a computational very efficient score- based test to investigate the association for a set of rare variants and ordinal traits. Through simulation, we evaluated the performance of our method. The proposed statistic obtain an accurate type I error rate, and it has higher power than many comparable binary trail methods. As a score-based test, our method can quickly calculate p-values, and so can easily be applied to genome-wide data sets of many hundreds of individuals. We apply this methodology to an epilepsy exon sequencing study to show the practical relevance of this approach.
|
