Pathway analysis and prediction analysis for various complex diseases
Principal Investigator: Professor Taesung Park
Approved Research ID: 58105
Approval date: June 24th 2020
Our research project has several interests. The first aim is using pathway level statistical models developed by our research group, to investigate the underlying mechanisms of complex traits. In particular, our group is interested in metabolic syndrome, type 2 diabetes, obesity, and hypertension. There are multiple rationales behind pathway analysis. A single biological function is rarely carried out by a single molecule but is constituted by a complex interaction of biological components in different pathways. Pathway-based analysis that can detect these relevant pathway is important to the understanding of the complexity of biological roles played by these components. Therefore, our group has developed a pathway-based method called Pathway-based approach using hierarchical components of collapsed rare variants(PHARAOH) for the pathway analysis. Recently, we extended this method to handle various types of data. These methods can identify associations between a trait and entire pathways and can also simultaneously quantify both the effects of the pathway and its components on the phenotype using a single model. The second aim is to build prediction models for complex traits such as Type 2 diabetes. Our group has developed several prediction models using various statistical methods for the Korean population. We would like to extend our models to other population groups using the UK Biobank data. All model building and evaluation processes will be performed for several traits of interests using both Korean and UK Biobank data. The following prediction models will also be considered such as logistic regression, support vector machine (SVM), and random forest (RF) as well as deep neural network (DNN) models. The above two interests will also be extended to the UK Biobank's various types of data. In summary, our research will identify pathway level candidates for further analysis to uncover biological mechanism of various traits. In addition, by providing effective analysis methods for various data, our research will also accelerate other ongoing studies with aims of discovering mechanisms for complex traits prediction. Furthermore, our group will build high-performance prediction models using statistical methods for various traits.