Association analysis of rare variants with sequencing data

点击翻译按钮获取中文摘要

An emerging and challenging research field in genetics is to detect associations between com- plex traits and rare variants (RVs) with next-generation sequencing and Exome Chip data. Due to ex- tremely low minor allele frequencies (MAFs) of RVs, many existing tests for common variants (CVs), such as the univariate test on each individual variant, most popular in genome-wide association studies (GWAS), may no longer be suitable. To boost power and facilitate biological interpretation, we propose combining information across multiple sources of data, which may or may not be of the same type. For the former, it leads to highly adaptive meta analysis suitable and powerful for com- bining multi-ethnic cohorts; for the latter, we integrate DNA genotype and sequencing data with gene networks, gene expression data and metabolomic data for association analysis of RVs. A common theme of the proposed methods is to explicitly account for genetic and phenotypic heterogeneity. For example, to account for genetic heterogeneity, we propose an adaptive network-based association test to aggregate information across multiple causal genes clustered in a network for a single cohort; for multiple cohorts, especially multi-ethnic ones, our proposed meta-analysis test is highly adaptive to heterogeneous and varying association patterns across cohorts (e.g. only few cohorts contain causal RVs) and among RVs. The developed methods will be applied to detect associations of RV- cardiovascular traits with the sequencing and other omic data from the ARIC study. We will develop and distribute software implementing the proposed methods. The proposed research is in line with the NHLBI's continuing interest in whole genome/exome sequencing and integrative omics analysis as evidenced by its TOPMed Program and NIH's other Precision Medicine initiatives.