Title page for ETD etd-07092007-142017

Type of Document

Thesis

Author

Joshi, Sonali

URN

etd-07092007-142017

Title

Estimating Selection Coefficient using the Ancestral Selection Graph

Degree

Master of Science

Department

Biological Science, Department of

Advisory Committee

Advisor Name	Title
Peter Beerli	Committee Chair
David Swofford	Committee Member
Gavin naylor	Committee Member

Keywords

Approximate Bayesian Computation
Coalescent
Summary Statistics
Selection Coefficient
Strength of Selection
Ancestral Selection Graph

Date of Defense

2007-06-01

Availability

unrestricted

Abstract

Detecting and measuring selection is of fundamental importance in many population genetic studies. The objective of this study is to estimate the strength of selection acting at a locus by studing polymorphism in neutral regions of DNA that are tightly linked to the given locus. Kingman’s Coalescent theory gives us a model to reconstruct the ancestral history of a sample for neutrally evolving sites. The Ancestral Selection Graph (ASG), introduced by Krone and Neuhauser is an extension of the neutral coalescent process and incorporates selection. However, it’s use has been limited due to computational issues in simulating the graph. In this study I use the Ancestral Selection Graph conditional on the sample allele configuration, as described by Slade for efficient simulation of selected genealogies. Inferences in the coalescent framework are often based on Likelihood or Bayesian approaches. As population genetic models get more complex there is a growing interest in simulation based Approximate Bayesian Methods. Tavare et al. first described an Approximate Bayesian Computation (ABC) method for simulating observations from posterior distributions without the use of Likelihoods. These methods use summary statistics to study the data and infer parameters. An estimator of Selection Coefficient is built using the Ancestral Selection Graph to simulate selected genealogies and the Approximate Bayesian approach to estimate parameters. The effect of the choice of summary statistics, priors and run lengths on estimation of parameters are explored
using simulated data.

Files

Filename		Size	Approximate Download Time (Hours:Minutes:Seconds)
Filename		Size	28.8 Modem	56K Modem	ISDN (64 Kb)	ISDN (128 Kb)	Higher-speed Access
	sj_thesis.pdf	650.75 Kb	00:03:00	00:01:32	00:01:21	00:00:40	00:00:03