Rare variant method for identity-by-descent detection in sequence data, and the time to most recent common ancestor given identity-by-descent segment length

Abstract

Identity-by-descent (IBD) is defined as two individuals sharing a haplotype they have both inherited from a common ancestor, where a haplotype is a segment of genetic material found on a single homologue. Existing methods for IBD detection were written to analyze low-density SNP arrays. These methods may not be well-suited to effectively report shared haplotype segments in sequence data because sequence data sets have an abundance of rare variants and an overall higher density of markers. We present a new method, RV-IBD, which uses the sharing of rare variants in sequence data to report IBD segments. Compared to existing methods for IBD detection, RV-IBD by itself did not show improved power and accuracy, but merging reported tracts from RV-IBD and the best-performing existing method did lead to improved power and accuracy. The RV-IBD method complements and is competitive with existing IBD detection methods using sequence data, showing that rare variants can be informative for IBD detection. We investigate the distribution of time to most recent common ancestor given the length of a shared IBD segment. In populations of constant size, we find the distribution to be Gamma with a scale parameter dependent on the length variable. In exponentially growing populations, we discuss trends in the conditional distributions.