Understanding the genetic architecture of traits-the frequencies, numbers, and effects of genetic variants that cause interpersonal differences-has been one of the central goals of statistical, molecular and evolutionary genetics over the last fify years. Twin/family studies have showed that most traits, including mental disorders, are highly heritable, recent genome-wide association studies (GWAS) have discovered thousands of single nucleotide polymorphisms (SNPs) reliably associated with these traits, and forthcoming whole-genome sequence data will allow a much more thorough investigation into genetic variants that underlie trait heritability. In the midst of this deluge of data, however, fundamenta questions about the genetic architecture of traits remain unanswered or are poorly characterized. Although twin/family studies have detailed the heritability of hundreds of traits, the degree to which this heritability is due to additive effects of genetic variants remains unclea. Although GWAS has demonstrated that a huge number of genetic variants must be responsible for trait heritability, the relative importance of common (shared by people worldwide) versus rare (specific to populations or extended families) genetic variants remains unclear. Finally, it is unclear whether genetic variants that predict traits in one ethnicity or population typically predit those same traits in other ethnicities or populations. As the field turns to whole-genome sequencing in the years ahead, it is crucial, now more than ever, to have a better understanding of these fundamental questions about the genetic architecture of traits. Doing so should help guide future analytic and investment decisions. We propose the development of methodologies that will help investigators greatly reduce the uncertainty surrounding the genetic architecture of traits using existing SNP data and, as it becomes available, sequence data. First, we demonstrate a method that allows the full additive genetic variation of a trait to be accurately estimated using simulated SNP data, and describe several advances that we will work on in order to make this method feasible to use on real SNP data. Second, we describe how sequence data can be used to accurately estimate the importance of common versus rare genetic variants, and propose the development of a method that will allow this approach to be used on existing SNP data. Third, we show a method that allows investigators to understand the degree to which SNPs that predict a trait in one ethnicity also predict that trait in another ethnicity, and we propose developing two extensions of this that (a) clarify why such differences occur and (b) make this approach applicable to understanding the specificity of SNP associations between subpopulations. Finally, we will apply these methods to the three largest case-control SNP datasets on Major Depressive Disorder, Bipolar Disorder, and Schizophrenia. By project's end, we anticipate having tools that allow for a much clearer understanding of the genetic architecture of these and other heritable phenotypes.
This project will refine and develop methodological tools that provide a much clearer understanding of three fundamental questions about heritable traits: (1) To what degree is the heritability of traits due to the effects of additive (and therefore more easily detected) genetic variants? (2) What is the relative importance of common (shared by people worldwide) versus rare (specific to populations or extended families) genetic variants? (3) To what degree are previous genetic discoveries, made mostly using samples of European descent, relevant to non- European populations?
Showing the most recent 10 out of 43 publications
