Statistical Methods For Genetic Epidemiology

我们现在正在考虑的一个特别重要的设计涉及一个“四分体”结构，除了两个亲本之外，还有一个受影响的后代和一个未受影响的后代。这个设计已经在Two Sister Study（由Susan G.科门的治疗），该研究正在评估遗传和环境风险因素在非典型性（50岁以下）乳腺癌中的共同作用。 不一致的同胞对允许估计暴露的影响，而嵌入式的情况下，父母的三联体允许检测单倍型，赋予保护或风险。一起分析的四分体应该提供一个强大的设计评估基因与环境的相互作用。我们一直致力于开发和评估用于四分体设计的方法。 两姐妹研究完成了核心家庭的登记，其中一个女儿在50岁之前患上乳腺癌，另一个女儿未受影响。这将在另一个项目下说明。 需要探索遗传基因型以及肿瘤特征，以研究预测治疗后临床病程的因素，并在此背景下开发改进的统计方法。 在NIEHS的IRP和Susan G.为了治疗科门，我们通过与约翰霍普金斯遗传疾病研究中心的合同，根据这些数据进行了一项全基因组关联研究，这些发现现已发表。我们还参加了GAME-ON联盟，该联盟提供了来自新开发的onco-chip的额外SNP。 我们还进行了基于风险评分的分析，该评分是基于先前在早期GWAS研究中重复的SNP开发的（综述中的论文）。 我们已经开始探索遗传风险评分和已知风险因素之间可能的相互作用。 在我们早期关于病例-父母三联体的工作的方法学扩展中，我们开发了一种可靠的方法来解释父母的表型，并将这些方法应用于我们的两姐妹研究，其中约20%的母亲也患有乳腺癌。 我们与生物统计学研究生Alison Wise一起开发了识别与风险相关的X染色体变异的方法。我们评估了我们的新方法的性能，将其应用到DbGap数据的出生缺陷，唇裂。 我们的方法，PIX-LRT，利用父母的信息在一个强大的方式，除了传输失真，从而使更有效地利用数据比现有的方法。艾莉森去年成功地通过了论文答辩。我们已经扩展了该方法，允许在X染色体上发现单倍型（多个SNP彼此接近）。 我们的方法来模拟情况下，父母的三元组与现实的LD结构已经尝试（使用dbGaP口裂数据）和测试，我们现在已经提交了一份文件描述它。软件实现模拟已经张贴，并正在免费提供。 基本方法可用于模拟数量性状数据或定性数据，无论是病例对照设计，病例父母设计，病例同胞设计或数量性状设计，包括基于结果的抽样的可能性。这项工作在今年8月于巴尔的摩举行的联合统计会议上提出。

A particularly important design we are now considering involves a "tetrad" structure, with one affected and one unaffected offspring, in addition to the two parents. This design has been implemented in the Two Sister Study (funded in part by Susan G. Komen for the Cure), which is assessing the joint role of genetic and environmental risk factors in young-onset (under age 50) breast cancer. The discordant sib pair allows estimation of effects of exposures, while the embedded case-parent triad allows detection of haplotypes that confer either protection or risk. The tetrad analyzed together should provide a powerful design for assessing gene-by-environment interaction. We have been working on developing and evaluating methods for use with the tetrad design. The Two Sister Study completed enrollment of nuclear families where one daughter developed breast cancer before age 50 and the other daughter is unaffected. This is described under a separate project. Inherited genotypes, together with tumor characteristics, will need to be explored to investigate factors that predict the clinical course following treatment, and improved statistical methods will also need to be developed in that context. With support from the IRP of NIEHS and Susan G. Komen for the Cure, we undertook a genome-wide association study based on these data through a contract with the Center for Inherited Disease Research at Johns Hopkins, and those findings have now been published. We also are participating in the GAME-ON consortium, which has provided additional SNPs from the newly developed onco-chip. We have also carried out analyses based on a risk score that was developed based on the SNPs that have previously been replicated in earlier GWAS investigations (paper in review). We have begun to explore possible interactions between the genetic risk score and known risk factors. In a methodologic extension to our earlier work on case-parent triads, we developed a robust method to account for parental phenotypes, and applied those methods to our Two Sister Study, in which some 20% of the mothers also had breast cancer. Together with a graduate student from UNC Biostatistics, Alison Wise, we have developed methods for identifying variants on the X chromosome related to risk. We assessed the performance of our new method by applying it to the DbGap data on the birth defect, oral cleft. Our method, the PIX-LRT, makes use of parental information in a robust way in addition to the transmission distortion, and thus makes more efficient use of the data than do existing methods. Alison successfully defended her thesis last year. We have extended the method allow discovery of haplotypes (multiple SNPs close to each other) on the X chromosome. Our method for simulating case-parent triads with realistic LD structure has been tried (using dbGaP oral cleft data) and tested and we have now submitted a paper describing it. Software for implementing simulations has been posted and is being made freely available. The basic approach can be used to simulate either quantitative trait data or qualitative data, either in a case-control design, a case-parents design, a case-sibling design or a quantitative trait design, including the possibility of outcome-based sampling. This work was presented at the Joint Statistical Meeting in Baltimore this August.