About
Scientific rationale: The arrival of novel biotechnologies provides new opportunities to facilitate genetic and functional genomics research for human disease. However, it is still a challenge to make complete use of the information embedded in genetic, medical image, and health informatics data to provide holistic biological insights. In addition, the complete pathological (disease-related) processes from molecules to complex diseases are generally not yet understood. To overcome these issues and effectively translate these diverse data to clinically useful knowledge, there is an imperative need to develop advanced and robust machine learning methods for investigating the genetic mechanisms of disease.
Aims: In this application, we specified the different aims to offer the most considerable flexibility in machine learning approaches for identifying disease-related genes or biological factors to better understand the mechanisms involved in the etiology of disease and to predict disease risk. First, we will present a machine learning framework to integrate different data sources with UK Biobank data to select disease-related genes/molecules. Second, we will propose a machine learning method to discover genetic effects and/or gene-environment interactions that are relevant for disease. Third, we will develop an advanced probability model to explore the regulatory effects between genes and prioritize causal gene candidates. Last, we will use machine learning strategies to develop disease prediction models which integrate various different types of biological data.
Project duration: The project period will be completed in 36 months.
Public health impact: Achieving these aims will enable us to discover novel disease-related genetic/environmental factors, and causal genes leading to complex diseases. Findings from the proposed work will be essential for revealing new insights into the complicated biological processes of complex diseases such as mental illness, obesity, and osteoporosis. This work will ultimately enrich current medical treatment strategies with efficient clinical prediction/prognosis/intervention models for improving health outcomes.