人类基因组的连锁不平衡强度变化及应用

【作者】 何云刚；

【导师】 金力；

【作者基本信息】 复旦大学 ， 群体遗传学， 2005， 博士

【摘要】 连锁不平衡是指基因组中不同位点间存在的非随机关联。它的存在，是进行基因型与表型间的关联分析和表型相关基因精细定位基本条件。单倍域的发现掀起了近年来连锁不平衡研究的热潮。仅用少数几个位点作为标签就可以有效地对单倍域内常见单倍型进行区分。使用单倍型标签位点进行分型工作可以节约分型成本并提高关联分析的统计效力。Hapmap项目使用三个重要的洲际群体产生大量高质量的分型数据。这些数据可以用于挑选标签位点。但是从这些群体挑选的标签位点是否可以用于其他人群而不带来大的统计效力损失和工作量的浪费仍然需要阐明。连锁不平衡受突变、重组、漂变、混合、以及其他一系列复杂的群体历史因素和一些随机因素影响。Wang等证明单倍域可以从有限大小的群体中自然产生。Zhang等通过模拟显示群体历史和重组热点可以影响单倍域的形成。McVean等从统计上推测基因组上广泛存在局部的重组率差异。而对芬兰群体和其他一系列隔离群体的连锁不平衡研究证明特殊的群体历史可以产生独特的连锁不平衡特征。深入了解当前人类基因组中的连锁不平衡产生和衰退的历史过程对连锁不平衡的应用具有指导意义。本研究通过对21号染色体上20，000多个单核苷酸多态性位点进行高质量的基因分型，考察了汉族、壮族、苗族、佤族和维吾尔族五个重要的东亚群体和非洲裔美国人，欧洲人，波利尼西亚的萨摩亚群体等三个非东亚群体的连锁不平衡特征。根据这些群体的21号染色体的位点间连锁不平衡状态的保守性，推测当前世界范围的非洲以外群体的连锁不平衡特征受这些群体分化前曾经经历的共同的瓶颈效应的严重影响。这一共同的瓶颈是非洲以外群体的位点间连锁不平衡状态具有较高的保守性的基础。通过比较不同群体的连锁不平衡的延伸长度与群体遗传学距离的相关性，可以看出陈旧的重组信息的丢失导致的连锁不平衡增强是最近十余万年，世界范围人群中连锁不平衡变化的主流。对比汉族，维吾尔族和欧洲人群体中位点频率的相互关系，我们得到维吾尔族同时具有东西方遗传成分的清晰证据。用同样的方法，本研究排除了萨摩亚土著群体与西方人存在明显的群体混合的可能。萨摩亚群体具有比较强的连锁不平衡。从连锁不平衡特征可以估计对连锁不平衡产生了重大影响的历史事件，大约发生在4500年前。对比经历过严重瓶颈的萨摩亚群体和其他群体间LD的保守性，我们更加确信非洲以外群体经历的共同的瓶颈是当前这些群体具有高的连锁不平衡程度的最主要原因。本研究通过直观有效的方法比较了被考察群体间标签位点的可移植性。结果显示标签位点可以在东亚群体间交换使用而不会带来大的统计效力损失和分型工作的浪费。同时论文还对不同情况下标签位点的选择策略作出了建议。更多还原

【Abstract】 Linkage disequilibrium (LD) is nonrandom association between different loci. It is part of the genome information on population level. Existence of LD between markers and suspect loci is the prerequisite for association analysis and fine mapping of disease suspect genes.According to discrete changes of LD in local region of human genome, loci in the small region could be divided into different haplotype blocks. Only few SNPs are needed to represent most haplotype diversity in each block. These haplotype tagging SNPs (htSNPs) could be used as markers in association study. Lower cost and higher statistic power are expected in this approach.Three continental populations had been chosen in Hapmap poject for high-quality genotyping. But, efficiency of using htSNPs from these populations in other ones still needs to be validated.Linkage disequilibrium was affected by mutation, recombination, drift, admixture as well as a series of other population demographic and random factors. Wang et al proved that block-liked structure could be generated naturally in a population with limited size. Zhang et al shows that population demographic events and recombination hotspot could effect the procedure of haplotype block genesis. McVean et al inferred that recombination rate is greatly different even in locally genome region. Research work about Finland and other isolated populations shows that special population history could produce special LD pattern in it. Understanding more about mechanism and procedure of LD reinforcement and decay will accelerate the using of LD in genetics research work.For exploring LD pattern in human genome, we chose more than 20,000 loci from human chromosome 21 and typed in 368 individual samples. These samples were come from five Asian populations (Han, Zhuang, Miao, Uyghur, Wa) and three other populations (European, African American, Samoan). Our result shows that LD patterns of non-African populations were largely affected by common bottleneck events in their history, which have contributed a lot to the conservative LD patterns among non-African populations.By comparing correlations between average LD extending length and genetic distance among different populations, we indicated that LD reinforcement generated by information loss of historical recombination events is the leading mode of theworldwide change of LD patterns among different populations in the last 100,000 years.Confident evidence from frequency relationship among Han, Uyghur and European populations clearly shows admixture of West and East in Uyghur population. By the same approach, we find no evidence of admixture between Samoa aborigines and Europeans.From pattern of LD decay in Samoan, we inferred that serious bottleneck events that contributed more to LD in Samoan were happened about 4500 years before. Result of LD sharing between Samoan and other population reinforced previous opinion, that seriously and enduringly common bottleneck events is mainly case of conservation of LD pattern among non-Afro populationsBy a direct and efficacious approach, our research compared the universality of tagging SNPs, which have warranted the using of tagging SNPs between East Asian populations without the remarkable loss of statistical power. Strategies for tagging SNP selection in association study are also discussed in this paper.更多还原