Statistical Power of Multivariate over Univariate Mixed Model Association Analysis
Aims: Develop an efficient and robust algorithm for disease traits of large-scale data from UK biobank!such as obesity, ADHD, autism, schizophrenia.
Scientific rationale: Analyzing multiple traits simultaneously can reveal more disease-related genes, reduce the number of performed statistical tests and alleviates the computational burden.
Project duration: The study will last 36 months. We first do computer programming and writing, then simulate 2, 3, 4, 5 and 6 correlated phenotypes to compare computational efficiencies, and 2 of which to demonstrate statistical property based on UK biobank data. Finally sort out the computer simulation results.
Public health impact: Using large scale data from UK biobank, we will create theoretical system of efficient mixed model association analysis, which involving various types of physical!physiological!biochemical!and disease traits. Compared to existing methods, our algorithms are able to find more genes and accurately predict complex disease with very high computational efficiency.