Bayesian Hierarchical Methods for Localized Analysis of Genic Intolerance to Variation

点击翻译按钮获取中文摘要

Project Summary: The goal of this proposed mentored research is to tie genetic variation to disease by analyzing regions that are intolerant to variation. Identifying regions that are intolerant to new variation can help localize regions of potential functional importance and biologic relevance. Large public population consortia are now accumulating datasets of sufficient size to detect regions subject to evolutionary selective pressures at an increasingly granular level. However, there remains a shortage of appropriate analytical tools that are built to specifically address important issues of disease heterogeneity across diverse populations. Despite the fact that clinical exome sequencing is increasingly used for improved diagnostic evaluation, many genetic disorders remain uncharacterized and diagnosis rates are still relatively low. In Aim 1, I will develop methodology that localizes regions intolerant to variation and differential isoform expression associated with disease. Many genes display tissue dependent transcript isoforms indicating potential functional implications of different isoforms. I will characterize selective pressure across all isoforms using Bayesian techniques by looking at patterns of genetic constraint across large standing populations, predominantly leveraging public data sets on the order of hundreds of thousands of samples. Then I will leverage existing expression data to isolate key isoforms across different cell and tissue types that are associated with diseases of interest. Then by accounting for regional intolerance to variation, a joint transcriptomic variation–intolerance approach can be employed to improve disease association testing. In Aim 2, I will analyze ancestry and cross species patterns of genetic intolerance to variation. The majority of genetic studies have focused on European populations, which ignores genetic and phenotypic diversity that can be leveraged to improve both targeted and overall diagnostic and clinical capabilities. I will test for ancestry and cross species patterns of genetic intolerance to variation and association with disease. Expanding to more populations will scale up the already large set of parameters being estimated; so, I will develop new statistical methods and software to improve optimization of parameter estimation for the Bayesian hierarchical models. I will isolate key ancestral populations with known differences in selective pressure to validate findings while then leveraging these new methods and population disease patterns further to isolate novel signals of ancestry-specific selective pressures. I will look for conserved regions across species to isolate essential exonic regions while also isolating unique regions in the context of human specific genetic variation and disease, such as neurodevelopmental disorders. During the training time for this proposed study I will focus on advancing my understanding of biologic mechanisms and clinical genetics to better inform the statistical genetics methods I develop. My mentorship and advisory committee consists of a strong multidisciplinary team of geneticists, pathologists, computational scientists, and biologists who will guide and collaborate with me to refine my work to improve variant interpretation and to advance precision medicine.