Midway evaluation in Statistics - Erna Valdís Ívarsdóttir

Title: Statistical methods in genome-wide association studies

Doctoral student: Erna Valdís Ívarsdóttir

Doctoral committee:

Daníel F. Guðbjartsson, VP of Applied Statistics at deCODE genetics

Gunnar Stefánsson, Professor at the Faculty of Physical Sciences, University of Iceland

Bjarni Halldórsson, Head of Sequence Analysis at deCODE genetics and Associate Professor at the department of Biomedical Engineering, Reykjavík University

Abstract:

Genome-wide association studies (GWASs) based on chips containing hundreds of thousands of single nucleotide polymorphisms have transformed the study of human genetics. Thousands of sequence variants associating with human traits and diseases have been discovered. Whole genome sequencing is now allowing the identification of most of the variation in the human sequence and associations with rarer mutations with greater effects. Currently, 28,075 Icelanders have had their whole genomes sequenced at deCODE genetics which has led to the identification of over 36 million sequence variants that have been imputed into 155,250 chip typed Icelanders.
Phenotypic data is available in the extensive deCODE database. A recent addition of phenotypes are being obtained in the deCODE health study, an ongoing population-based study, which involves a comprehensive phenotyping of the recruited subjects. A part of the collected measurements are audiometric tests and various ocular measures which have not been analyzed before in large GWASs.
Most GWASs to date has focused on discovering sequence variants that have an effect on the mean of traits or disease risk. However, variants can also affect the variability of traits, both within-subject variance and between-subject variance. Assessing these variance effects can potentially lead to the discovery of novel sequence variants but also has the potential of improving our understanding of previously associated sequence variants. 

The main goals of this project are:
(i) Estimating the effect of known glucose variants on the variance of glucose levels and examine the relationship between their genetic effect on glucose levels and their genetic effect on T2D. We also estimate how the effect of these variants on variance affect heritability estimates.
(ii) Using different statistical models to search for sequence variants associating with sensory measurements (auditory and ocular) obtained in the deCODE health study and explore the causal relationship between correlated traits and diseases.