Scientific rationale: Even though understanding how DNA sequences affect disease risk is a central problem in medicine, the knowledge for the genetic basis of complex diseases is still limited. On the other hand, integrative analysis of multiple genetic and genomic datasets turns out to be a beneficial method to gain new insights into the genetic mechanisms of complex traits. While appealing, the techniques used for integrative analysis (especially these tailored for UK Biobank data) are still primitive, and some new statistical methods are urgently needed.
Aims: In this proposal, we will develop integrative analysis methods that integrate UK Biobank data with other genetic and genomic datasets. Specifically, we plan to achieve with three related sub-aims. First, we will develop a deep learning/machine learning framework to improve the disease risk prediction accuracy. Second, we will propose a new method to detect how genes and environmental factors such as smoking status interact with each other. Third, we will propose new methods to identify and prioritize putative causal genes that have a direct effect on complex diseases. In the end, we will release public-domain software and online manuals.
Project duration: The project period will be maximally 36 months.
Public health impact: The proposed research might potentially identify some putative causal genes for complex diseases, significant gene-environmental interactions, and a new way to predict the risk of complex diseases. All these findings will help us gain insights into the genetic mechanisms of complex diseases and develop new prevention and diagnosis methods for complex diseases. The proposed research is in line with the UK Biobank strong interest in improving the prevention and diagnosis of complex diseases, including depression, schizophrenia, and Alzheimer’s.