Statistical Methods for RNA-seq Data Analysis

点击翻译按钮获取中文摘要

Project Summary Traditional RNA-seq studies collect RNA-seq data from bulk samples (bulk RNA-seq) and thus aggregate the signals from multiple cell types. Gene expression variation across samples may be due to difference of cell type composition or cell type-specific gene expression, and bulk RNA-seq data cannot distinguish these two factors. In fact, cell type-specific signals may be masked or even misrepresented by bulk RNA-seq data. Single cell RNA-sequencing (scRNA-seq) may overcome part of the limitations of bulk RNA-seq. However, in a foreseeable future, it cannot be applied to a large cohort due to cost and logistical barriers. In this R01 proposal, we propose new statistical/computational methods to study cell type composition or cell type-specific gene expression using bulk RNA-seq data, scRNA-seq data, or both bulk RNA-seq and scRNA-seq data. This approach can effectively exploit the huge amount of existing bulk RNA-seq data, and it can bring paradigm- shifting changes to many fields, for example, identifying cell types associated with a disease trait or defining new biomarkers using cell type-specific gene expression. We plan to achieve the following three specific aims. In Aim 1, we propose novel methods for cell type-specific differential expression analysis as well as methods to assess the association between cell type composition and covariates of interest. In Aim 2, we focus on the association between cell type-specific gene expression and germline genetic variants, i.e., studying cell type- specific gene expression quantitative trait loci (eQTLs). In Aim 3, we study the association between somatic mutations and cell type composition or cell type-specific gene expression.