Detecting ancient balancing selection : methods and application to human / Katherine M. Siewert.

Format:

Book

Manuscript

Thesis/Dissertation

Author/Creator:

Siewert, Katherine M., author.

Contributor:

Voight, Benjamin F., degree supervisor.

Tishkoff, Sarah, degree committee member.

Kim, Junhyong, degree committee member.

Messer, Philipp W., degree committee member.

Devoto, M. (Marcella), degree committee member.

University of Pennsylvania. Department of Genomics and Computational Biology, degree granting institution.

Language:

English

Subjects (All):

Penn dissertations--Genomics and computational biology.

Genomics and computational biology--Penn dissertations.

Local Subjects:

Penn dissertations--Genomics and computational biology.

Genomics and computational biology--Penn dissertations.

Physical Description:

xi, 107 leaves : color illustrations ; 29 cm

Production:

[Philadelphia, Pennsylvania] : University of Pennsylvania, 2019.

Summary:

Balancing selection can maintain genetic variation in a population over long evolutionary time periods. Identifying genomic loci under this type of selection not only elucidates selective pressures and adaptations but can also help interpret common genetic variation contributing to disease. Summary statistics which capture signatures in the site frequency spectrum are frequently used to scan the genome to detect loci showing evidence of balancing selection. However, these approaches have limited power because they rely on imprecise signatures such as a general excess of heterozygosity or number of genetic variants. A second class of statistics, based on likelihoods, have higher power but are often computationally prohibitive. In addition, a majority of methods in both classes require a high-quality sequenced outgroup, which is unavailable for many species of interest. Therefore, there is a need for a well-powered and widely-applicable statistical approach to detect balancing selection. Theory suggests that long-term balancing selection will result in a genealogy with very long internal branches. In this thesis, I show that this leads to a precise signature: an excess of genetic variants at near identical allele frequencies to one another. We have developed novel summary statistics to detect this signature of balancing selection, termed the [beta] statistics. Using simulations, we show that these statistics are not only computationally light but also have high power even if an outgroup is unavailable. We have derived the variance of these statistics, allowing proper comparison of [beta] values across sample sizes, mutation rates, and allele frequencies--variables not fully accounted for by many previous methods. We scanned the 1000 Genomes Project data with [beta] to find balanced loci in humans. Here, I report multiple balanced haplotypes that are strongly linked to both association signals for complex traits and regulatory variants, indicating balancing selection may be affecting complex trait architecture. Due to their high power and wide applicability, the [beta] statistics enable evolutionary biologists to detect targets of balancing selection in a range of species and with a degree of specificity previously unattainable.

Notes:

Ph. D. University of Pennsylvania 2019.

Department: Genomics and Computational Biology.

Supervisor: Benjamin F. Voight.

Includes bibliographical references.

Other Format:

Online version: Siewert, Katherine M. Detecting ancient balancing selection.

OCLC:

1132292176