ASSOCIATION BETWEEN PHENOTYPES AND DISEQUALIBRIATE LOCI

描述（由申请人提供）：已知的人类多态性的数量正在迅速增长，并且用于确定基因型的时间和金钱成本已大幅下降。这一结果可能会推动对影响人类表型的基因的研究，对理解、避免和治疗疾病产生积极影响。这也将影响对药物反应的理解和避免对药物的过敏反应。虽然数据收集的速度有了很大的进步，但处理如此庞大和复杂的数据集的统计方法却没有跟上。最紧迫的问题是如何测试表型和大量潜在相关基因座之间的关联。目前大多数方法需要多次测试，并遭受相关的统计能力损失。这个问题进一步复杂化，因为位点以及因此测试不是独立的，这导致过度校正。 连锁不平衡、单倍型块和热点的成对测量不足以描述近端位点等位基因状态之间的关联模式。更复杂的结合和群体遗传模型具有易处理性或完全结合来自单倍型样本的信息的问题。 图形建模提供了一个统计框架来精确地描述这类复杂的随机数据。这种经验方法可以提供简洁，准确和易于处理的代表性的联合分布的等位基因在近端基因座。这是直接相关的问题，如检测与表型变量的关联和选择信息子集的基因座。这种方法的巨大潜力在于，分类表型可以被包括在同一分析中，并且与多态性的关联与基因座间关联联合评估。 该建议是将研究人员已经开发的用于单倍体数据的图形建模方法扩展到二倍体数据、更大的基因组区域、混合种群和家族数据。

DESCRIPTION (provided by applicant): The number of known human polymorphisms is growing rapidly and the cost, both in time and money, for ascertaining genotypes has decreased substantially. The result can potentially drive forward the search for genes affecting human phenotypes with positive consequences for understanding, avoiding and treating disease. It will also impact the understanding of drug response and the avoidance of allergic reactions to drugs. While the rate of data gathering has advanced so significantly, the statistical methods to deal with such large and complex data sets have not kept pace. The most pressing problem is that of how to test for association between phenotypes and a vast array of potentially associated loci. Most current methods require multiple testing and suffer the associated loss of statistical power. This problem is further complicated as the loci, and hence the tests, are not independent which leads to over correction. Pairwise measures of linkage disequilibrium, haplotype blocks and hot spots inadequately describe the patterns of associations between the allelic states at proximal loci. More sophisticated coalescent and population genetic models have problems of tractability or of fully incorporating the information from haplotype samples. Graphical modeling provides a statistical framework for characterizing precisely this sort of complex stochastic data. This empirical approach can provide concise, accurate and tractable representations of the joint distribution of alleles at proximal loci. This is directly relevant to such problems as detecting association with phenotypic variables and selecting informative subsets of loci. The great potential of this approach is that categorical phenotypes can be included in the same analysis and association with polymorphisms assessed jointly with the inter locus associations. This proposal is to extend graphical modeling methods already developed by the investigators for haploid data to diploid data, larger genomic regions, admixed populations and family data.