Investigations of methods for integrative causal inference modelling
Principal Investigator: Dr Celia Greenwood
Approved Research ID: 27505
Approval date: July 1st 2017
We are interested in evaluating existing statistical analysis methods for investigating when genetic variants have effects on more than one trait or phenotype, and in developing improved statistical methods of analysis. In this context, we would like to look at the performance of methods using both common and rare genetic variants. Improved methods for analyzing pleiotropy and causal relations will benefit those who are trying to understand the causal relationships between genetics and phenotypes. The methods will be applicable to many researchers' analyses. Since this is methodological work, we will use the data to evaluate the use of the methods in situations that have been well characterized in the literature. For example, we propose to examine SNP associations on lipid traits and relationships with cardiovascular disease. For assessing performance of causal inference methods including bias and power, we propose to divide the data set randomly into partitions to look at stochasticity of the results. We would be interested in two different subsets. The full data on a restricted set of genotypes for a specific set of lipid-related phenotypes, and a larger set of genotypes on individuals where have been measured. Another researcher (and collaborator) from our institution has requested and obtained UK biobank data (Dr. Brent Richards), so if were possible to use the genetic data that have already been downloaded, that would be ideal. i.e. link our applications.