Skip navigation
Please use this identifier to cite or link to this item:
Title: Detecting, Modelling, and Interpreting Gene-by-Environment Interactions
Authors: Nelson, Emily Spencer
Advisors: Storey, John D
Contributors: Quantitative Computational Biology Department
Keywords: gene-by-environment
Subjects: Bioinformatics
Issue Date: 2017
Publisher: Princeton, NJ : Princeton University
Abstract: A complete understanding of the relationship between genotypes and complex phenotypes requires careful study of gene-by-environment interactions (GxE). One must not, however, conflate GxE interaction with statistical interaction between the genotype and the environment. In this thesis, I demonstrate the flaw in equating statistical interaction with GxE interaction, namely that statistical interaction is inherently scale-dependent, and propose a stricter definition of GxE interaction more consistent with the biological interpretation. Using both real and simulated gene expression data, I show that my stricter definition (implemented in a procedure termed multiple environment self-association, or MESA) not only performs better, but also solves the problem of scale-dependence when quantifying GxE interaction. I extend the MESA procedure to the Gene-Tissue Expression (GTEx) dataset, a large tissue-specific RNA-Seq experiment in humans. This data allows me to illustrate one of the great strengths of MESA: the ability to easily extend the procedure to an arbitrary number of tissues or environments without loss of power or an increase in model complexity. With this data, I show that patterns of self-association between tissues vary dramatically, suggesting that GxE interactions in different tissues may have a strong effect on phenotype. I also discover some interesting connections to biological processes and diseases. In the second half of this thesis, I extend my study of GxE interactions by examining how such interactions change and develop over time, with the use of a yeast experiment which provides fitness as well as gene expression phenotypes. I marshal this data through a complete bioinformatics pipeline, including a novel normalization procedure and model selection process. Contrary to previous results in yeast, I show that while complex patterns of change over time across conditions play a large role in gene expression phenotypes, it appears that marginal interaction between the allele and condition do not. However, GxE interaction does appear to play a role in fitness, and I additionally observe interactions between loci that seem to affect fitness. I propose a hypothesis explaining the dearth of GxE interactions observed in this experiment versus previous ones, and also shed light on some other interesting biological effects at play.
Alternate format: The Mudd Manuscript Library retains one bound copy of each dissertation. Search for these copies in the library's main catalog:
Type of Material: Academic dissertations (Ph.D.)
Language: en
Appears in Collections:Quantitative Computational Biology

Files in This Item:
File Description SizeFormat 
Nelson_princeton_0181D_12293.pdf10.11 MBAdobe PDFView/Download

Items in Dataspace are protected by copyright, with all rights reserved, unless otherwise indicated.