罕見變異關聯性分析的分類與介紹

Abstract

科學家們發現多個罕見變異比單一常見變異更能偵測出遺傳性疾病的發生，而近年來次世代定序的發展，促使了罕見變異關聯性分析的進步，也因此發展出了各式各樣的分析方法。本文將把眾多方法加以分類和介紹，並使用Genetic Analysis Workshop 18 (GAW18) 的全基因組資料及其模擬表現型資料配合各個方法進行分析。大多數的方法樣本間需要互相獨立，因此我們僅使用其中不相關的 142 位的資料，並由 dbSNP 的版本 GRC37 中整理出第三條染色體上的 1226 個基因區域，再使用各個方法的軟體或 R 的函數進行分析，最後根據 GAW18 提供的模擬答案比較各個方法的分析結果。

Scientists find that multiple rare variants are more relevant to hereditary disease than single common variant. Recently, the development of next-generation sequencing (NGS) prompts the progress of rare variant association study. A wide variety of statistical methods for relating rare variants to phenotypes exist at present. This thesis aims to classify these methods and introduce some of them. We use whole genome sequence data and simulated phenotype data from Genetic Analysis Workshop 18 (GAW18) to illustrate these methods. Because most methods assume the samples are independent, we only analyze 142 independent samples in the dataset. We sort out 1226 genes on chromosome 3 based on the database of dbSNP’s edition GRC37, and implement each method’s software or R function to perform analysis. Results are compared with the answer provided by GAW18.