Events and Meetings of Italian Statistical Society, Advances in Latent Variables - Methods, Models and Applications

Font Size: 
Genetic Consortia, Risk Scores and Mendelian Randomization
John Thompson

Last modified: 2013-06-14


Genetic consortia that meta-analyse the results from genomewide association studies have been very successful at identifying the gene related to a wide range of diseases and traits. It is interesting to ask whether a genetic risk score that predicts the risk of one trait will also predict the risk of another, since this would suggest a common causal mechanism. A more formal method of addressing the same issue is to perform a Mendelian Randomization in which the genes act as instrumental variables in a regression of one trait on the other. We consider the circumstances under which these two approaches are equivalent and consider practical issues such as how to choose the genes that are used. The methods will be illustrated by using data on the association between heart rate and coronary artery disease.

Full Text: PDF