Using Rcpp to speed up tool for controlling for multiple testing in genetic studies

Presented by:
Zuofu Huang (Macalester College)

For many years, humans have been striving to understand genetic causes of diseases. Unfortunately, the large majority of genetic studies have largely focused on populations of Europeans ancestry; populations with a more diverse genetic ancestry such as Hispanics/Latinos and African ancestries are largely underrepresented. Admixture mapping is a powerful tool for uncovering the relationship between one’s disease status and genetics in populations with mixed ancestry. In standard admixture mapping studies, we use a technique know as marginal regression and step along the genome looking for associations between genetic material and disease status at each locus. Given the large number of hypothesis tests being conducted (hundreds of thousands to millions), we are more prone to making false discoveries. In recent years, tools for reducing multiple testing errors in admixture mapping studies have been proposed. This summer, I used Rcpp to improve on an existing tool for reducing multiple testing errors in admixture mapping studies; improvements show an up to 78% increase in speed without a loss in accuracy.