Bayesian AnaLysis of Diabetes for Enhanced biomarkeR and drug target

Type 2 diabetes mellitus (T2DM) is a common disease with a prevalence in Denmark of 250.000 diagnosed patients. T2DM has a strong underlying genetic component; for example, having a parent with T2DM, your risk is approximately 40%. Genetic variants associated with T2DM currently only explain 3% of the disease risk. Increased power to detect variants is mainly obtained by increasing sample size, or as we propose better use of existing data. We will increase detection power by developing a multi-trait and multi-component Bayesian Linear Regression (MT-BLR) model that combine information on multiple correlated traits and information on groups of genetic variants located within functional units (e.g., pathways). Our modelling approach allow better use of existing data such as functional marker information in biological databases, and availability of large independently collected genotype and phenotype data sets for a range of diseases, including T2DM. We will use these existing data to develop statistical models that better use information on correlated traits and disease for detecting genetic signals underlying T2DM. We anticipate that our modelling approach will generate novel biological insight into T2DM disease aetiology. Importantly, the discovery of novel genetic markers associated with T2DM can help identify potential drug targets for T2DM including genes and regulatory elements located nearby disease-associated variants or biological pathway. We will gather this information in a database for T2DM for genomic informed drug target identification, and the corresponding statistical methodologies will be implemented in open-source software packages allowing us to share openly with other research groups in academia and pharmaceutical industry. While the focus in this project is on improving the drug target discovery process our modelling approach can also be used for developing more accurate genetic risk predictors for complex diseases. We are convinced that our novel statistical modelling approaches and genomic informed drug target identification strategy can be applied to other complex diseases (e.g., cardiovascular disease) contributing to the identification of potential drug targets for a range complex diseases.