Bayesian Modeling of Spatial Ancestry Mapping

• Activity Number: 608
• Type: Contributed
• Date/Time: Thursday, August 7, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Kaustubh Adhikari*+ and Andres Ruiz-Linares
• Companies: and University College London
• Keywords: genetic ancestry ; spatial maps ; bayesian modeling

Abstract

In large-scale human genetic studies such as Genome-Wide Association Studies, we are increasingly working with mixed-ancestry populations, and we need to get an idea of their genetic ancestry before any analysis, as population structure is one of the major confounders. Birthplace is another confounder as geographic distance creates both genetic isolation and environmental variation. This makes us curious about spatial ancestry maps across a region. But a common problem is the tendency of agglomeration in the cities, and to address it, studies often collect birthplace data for parents and grandparents. In this study we provide a Bayesian model to produce spatial ancestry maps taking into account the birthplaces of an individual up to grandparents. We allow for recent immigration from outside, as that can distort the genetic makeup of a population. The advantage of Bayesian modeling is that it allows for missing data, which is typical as one goes deeper along the family tree. Another flexibility of this method is that it can have any number of ancestry components in the model. We demonstrate it on a study of Latin America with estimated European, Amerindian and African ancestry.