針對目標基因外顯子區域探討次世代基因定序偵測拷貝數變異的方法

Abstract

對於疾病變異(disease-variants)的偵測來說，結合次世代基因定序(next-generation sequencing) 技術的外顯子組測序 (exome sequencing )是一種極具成本效益之方法。其中，找出基因拷貝數變異（copy number variations , 簡稱 CNVs）即是外顯子組測序的重要應用之一。近年來已發展出許多偵測外顯子組拷貝數變異(exome CNV)的軟體，但這些軟體的性能和精確度卻未曾得到徹底地評估與比較。本篇論文將回顧四種較流行的外顯子組拷貝數變軟體之理論（CoNIFER、 cn.MOPS、exomeCopy和ExomeDepth），並利用UK10K計畫(http://www.uk10k.org/) 下的資料(蘇格蘭的214位精神分裂症患者)來對這四種軟體進行評估與比較。我們發現這些外顯子組拷貝數變異分析軟體，雖然有些流程相似，但仍因分析方法不同，導致最終偵測出來的拷貝數變異會有不少的差異。最後我們進一步分析軟體間偵測出來的拷貝數變異的差異，例如：一致性(concordance)、靈敏度(sensitivity) 和偏誤(bias)，進而有系統地評價這四個軟體的表現及給予一些建議作為未來研究的參考。

Exome sequencing using next-generation sequencing technologies is a cost-efficient approach for detection of disease variants. One of the important applications of exome sequencing data is to identify copy number variations (CNVs). During the last several years, there have been many exome CNV softwares developed , but the accuracy and performance of these programs have not been thoroughly evaluated. In this thesis, we will review the theory of four popular exome CNV softwares (exomeCopy, ExomeDepth, cn.MOPS and CoNIFER) and uses the data of UK10K project (http://www.uk10k.org/) (214 samples with schizophrenia in Scotland) to systematically compare and evaluate four softwares. Although some preprocess of four software are equal, the results of detecting copy number variation are still different due to some different analysis methods. We further calculate the concordance, sensitivity and bias of these four software to systematically evaluate their performance and give some suggestions for the usage of them.