對NIMH之精神分裂症資料作全基因相關性分析－以潛在類別方法將病人分組

Abstract

精神分裂症是一種複雜的疾病，同時受到基因遺傳及環境因素的影響。我們從NIMH取得case-control資料，其中包含741位精神分裂症患者和751位一般人，透過全基因相關性研究，希望可以找出導致精神分裂症的基因位置。因為不同症狀的精神分裂症導致的基因位置可能不同，所以我們先以潛在類別方法依正性負性症狀評量(PANSS)將病人分成數種症狀。又為了避免不同種族造成的影響，在作關聯性檢定時，用主成分分析調整種族。另外藉由基因型設算可以估計出未觀察到的基因。本文主要在探討以潛在類別方法將病人分組和用主成分分析調整種族的部分。

Schizophrenia is a complex disease caused by both genetic inheritance and environment factors. We use the case-control genome-wide association study (GWAS) with 741 cases and 751 controls from the National Institute of Mental Health (NIMH) database to test SNP association with schizophrenia. Because different symptoms of schizophrenia may be caused by different genes, we use the regression extension of latent class analysis (RLCA) to classify patients based on their responses to Positive and Negative Syndrome Scale (PANSS) measures, and perform GWAS separately for each individual patient subgroup. To avoid potential population stratification, we apply principal component analysis (PCA) to adjust for the race effect when testing the association. We also extract the genotypes of rare SNPs by imputation to evaluate the contribution to the disease from multiple rare variants. In this thesis, we focus on schizophrenia patient subgroup identification by RLCA and population stratification adjustment by PCA.