Improving Modeling and Prediction of Complex Traits and Diseases by Accounting for Context-Dependent Effects
Approved Research ID: 62347
Approval date: January 20th 2021
The aim of this 36-month-long project is to improve the accuracy of prediction models of complex human traits and disease. We call a trait or disease "complex" when it is affected by both genetics (i.e., genotypes) and environmental conditions such as smoking status, diet, and physical activity. Some examples of complex diseases are cardiovascular disease and diabetes. Being able to predict complex traits and disease risk from genotypes using statistical models is the ultimate goal of precision medicine. This will allow individuals at risk to be identified early to be able to take preventive actions. However, for the vast majority of complex traits and diseases, prediction accuracy achieved by currently available statistical models is low. We hypothesize that this low prediction accuracy is, at least partly, due to complex interactions between the genetic and environmental components that are not accounted for in most prediction methods currently available. Thus, we plan to fill this gap by investigating the role of gene-environment interactions and their inclusion into phenotypic prediction models for a variety of human complex traits and diseases.
The successful completion of this project will provide a thorough understanding of the portance of such interactions on a variety of human complex traits and diseases within the UK Biobank. This new knowledge will also result in improved prediction accuracy for such traits and will be a significant advance towards effective personalized medicine.