四种肿瘤体细胞单核苷酸突变检测方法的比较  

Comparison of Four Methods for Detecting Somatic Single Nucleotide Variant in Tumor Cells

作　　者：李晓东[1,2] 何小雨 陈玮 李瑞琳 赵丹[1,2] 祝海栋 张裕 代闯闯 陆忠华 迟学斌 牛北方[1,4] 郎显宇[1] Li Xiaodong;He Xiaoyu;Chen Wei;Li Ruilin;Zhao Dan;Zhu Haidong;Zhang Yu;Dai Chuangchuang;Lu Zhonghua;Chi Xuebin;Niu Beifang;Lang Xianyu(Computer Network Information Center, Chinese Academy of Sciences, Beijing 100190, China;University of Chinese Academy of Sciences, Beijing 100190, China;Center of Scientific Computing Applications ＆ Research, Chinese Academy of Sciences, Beijing 100190, China;Guizhou University School of Medicine, Guiyang, Guizhou 550025, China)

机构地区：[1]中国科学院计算机网络信息中心,北京100190 [2]中国科学院大学,北京100049 [3]中国科学院计算科学应用中心,北京100190 [4]贵州大学医学院,贵州贵阳550025

出　　处：《科研信息化技术与应用》2017年第6期77-87,共11页E-science Technology & Application

基　　金：基于国家高性能计算环境的教育实践平台(2016YFB0201900);国家自然科学基金青年基金(61702476);国家自然科学基金面上项目(31771466);青海省科技成果转化专项(2016-SF-127);生态环境损害鉴定评估平台构建及应用示范(2016YFC0503607);个性化药物-基于疾病分子分型的惠普新药研发(XDA12000000)

摘　　要：随着高通量测序成本的不断降低,基于DNA测序技术的肿瘤基因组研究已经成为揭示肿瘤分子机制的主流方法,并在临床诊断和治疗中逐渐得到应用。肿瘤体细胞单核苷酸突变变异(single nucleotide variant,SNV)作为最简单的一种基因变异类型,其检测会受到家系多态性、肿瘤异质性、测序和分析误差等多个因素的影响,从而导致一些假阳性的结果。目前,已有一些基于肿瘤基因组测序数据的体细胞SNV检测软件,如Varscan2,Mutect2,Strelka,Somatic Sniper等。本文选取四种典型的检测方法,对每种方法的检测原理进行研究,并使用ICGC-TCGA提供的全基因组数据,对上述四种变异检测软件进行测试。参照每种方法的分析流程,获得每种方法识别的候选变异位点集,并与真实的变异位点集合进行比较,分析每种算法的优缺点,从而为研究人员使用这些方法提供指导。With the cost of high-throughput sequencing continuous reducing, the research of tumor genome based on DNA sequencing technology has become the mainstream method to reveal the molecular mechanism of tumor, and it has gradually been applied in clinical diagnosis and treatment. The single nucleotide variation（SNV） of tumor somatic cells is the most common kind of genetic variations. Its detection will be affected by many factors such as family polymorphism, tumor heterogeneity, sequencing and analysis errors which will lead to some false positive results. There are some somatic cell SNV detection software based on tumor genome sequencing data now, such as Varscan2, Mutect2, Strelka, Somatic Sniper and so on. In this paper, we selected four typical detection methods are selected, and studied the detection principle of each method. We also used the whole genome data provided by ICGC-TCGA to test the four mutation detection software. With reference to the analysis flow of each method, we obtained four sets of candidate mutation sites identified by each method. Then we compared each set with the set of true mutation sites, analyzed the advantages and disadvantages of each algorithm. After our work, we concluded some suggestions that can provide to researcher when they want to use these methods.

关 键 词：体细胞单核苷酸变异 基因序列 突变检测 假阳性 测序深度 

分 类 号：Q524[生物学—生物化学]