Schork Nicholas J, Wessel Jennifer, Malo Nathalie
Department of Psychiatry, University of California at San Diego, La Jolla, CA 92093, USA.
Adv Genet. 2008;60:195-217. doi: 10.1016/S0065-2660(07)00409-9.
The availability of cost-effective, high-throughput genotyping technologies has generated a tremendous amount of interest in genetic association studies. This interest has led to the belief that one could possibly test thousands to millions of representative polymorphic sites on the genome for association with a trait or disease in order to identify the few sites that may be of relevance to the expression of that trait or disease. The choice of which polymorphic sites are "representative" and to be interrogated in such studies is problematic and has involved considerations of the putative functional significance of the sites as well as the linkage disequilibrium relationships between variations at those sites and other neighboring sites. We consider an obvious alternative to genotyping-based strategies and settings for association studies for which decisions about which variations to interrogate are obviated. Essentially, we anticipate a time when cost-effective, high-throughput DNA sequencing technologies are available and researchers will have actual sequence information on the individuals under study rather than information about what variations they possess at a few well-chosen polymorphic genomic sites. We consider Multivariate Distance Matrix Regression analysis to evaluate associations between DNA sequence information and quantitative traits such as blood pressure and cholesterol level. We evaluate the potential of the method in a few (albeit contrived) settings via simulation studies. Ultimately, we show that the procedure has promise and argue that consideration of DNA sequence-based association data should usher in a new era in genetic association study designs and methodologies.
具有成本效益的高通量基因分型技术的出现，引发了人们对基因关联研究的极大兴趣。这种兴趣导致人们相信，有可能在基因组上对数以千计到数百万个具有代表性的多态性位点进行与某一性状或疾病的关联性测试，以便找出可能与该性状或疾病表达相关的少数位点。在这类研究中，选择哪些多态性位点具有“代表性”并进行检测是个难题，这涉及到对这些位点假定的功能重要性的考量，以及这些位点与其他相邻位点的变异之间的连锁不平衡关系。我们考虑了一种明显的替代基于基因分型策略和关联研究设置的方法，在这种方法中，关于检测哪些变异的决策变得不再必要。从本质上讲，我们预计会有这样一个时代，那时具有成本效益的高通量DNA测序技术将可用，研究人员将拥有所研究个体的实际序列信息，而不是关于他们在少数精心挑选的多态性基因组位点上拥有哪些变异的信息。我们考虑使用多变量距离矩阵回归分析来评估DNA序列信息与诸如血压和胆固醇水平等定量性状之间的关联性。我们通过模拟研究在一些（尽管是人为设定的）情况下评估了该方法的潜力。最终，我们表明该方法具有前景，并认为对基于DNA序列的关联数据的考虑应该会开创基因关联研究设计和方法的新时代。
