Computational methods for high-throughput pooled genetic experiments

Abstract

Advances in high-throughput DNA sequencing have created new avenues of attack for classical genetics problems. This thesis develops and applies principled methods for analyzing DNA sequencing data from multiple pools of individual genomes. Theoretical expectations under several genetic models are used to inform specific experimental designs and guide the allocation of experimental resources. A computational framework is developed for analyzing and accurately extracting informative data from DNA sequencing reads obtained from pools of individuals. A series of statistical tests are proposed in order to detect nonrandom associations in pooled data, including a novel approach based on hidden Markov models that optimally shares data across genomic locations. The methods are applied to new and existing datasets and improve on the resolution of published methods, frequently obtaining single-gene accuracy.