Statistical Methods in Chronic Disease Research

描述（由申请人提供）：本研究的广泛，长期目标是开发创新和高影响力的统计方法，用于慢性病研究的设计和分析，重点是基因组学。这种竞争性更新应用的具体目标包括：（1）一般两阶段研究的有效估计，其中在第一阶段中对所有研究受试者测量可能不完整的多变量结果和廉价协变量，并且第一阶段信息用于最佳选择受试者以在第二阶段中测量昂贵协变量;（2）当在SNP阵列上对所有研究对象进行基因分型但仅对一小部分进行测序时，或者当直接关注未观察到的等位基因特异性拷贝数时，有效且高效地分析遗传关联;（3）随机效应模型下的荟萃分析，研究数量相对于研究样本量和变量选择而言较少，基于多种模型结构下多项研究的汇总统计量。所有这些问题都是由主要研究者的应用研究经验所激发的，并且与当前的基因组研究高度相关。所提出的解决方案是基于可能性和其他合理的统计原则。新方法的大样本性质将通过现代经验过程理论和半参数有效性理论严格地建立。将开发有效和稳定的数值算法来实现推理程序。所提出的方法将通过模拟真实的数据的模拟研究进行广泛的评估，并应用于几个主要的基因组研究，其中大部分是在国家生物技术研究所进行的。将免费提供附有适当文件的高效、可靠和方便用户的软件。这项研究不仅将推动生物统计学和统计遗传学领域的发展，而且还将影响美国癌症研究所和其他地方的慢性病研究。

DESCRIPTION (provided by applicant): The broad, long-term objectives of this research are the developments of innovative and high-impact statistical methods for the designs and analysis of chronic disease studies, with an emphasis on genomics. The specific aims of this competing renewal application include: (1) efficient estimation for general two-phase studies, in which possibly incomplete multivariate outcomes and inexpensive covariates are measured on all study subjects in the first phase and the first-phase information is used to optimally select subjects for measurements of expensive covariates in the second phase; (2) valid and efficient analysis of genetic association when all study subjects are genotyped on a SNP array but only a small subset is sequenced or when unobserved allele-specific copy numbers are of direct interest; (3) meta-analysis under random-effects models when the number of studies is small relative to study sample sizes and variable selection based on summary statistics of multiple studies under a variety of model structures. All these problems are motivated by the principal investigator's applied research experiences and are highly relevant to current genomic studies. The proposed solutions are based on likelihood and other sound statistical principles. The large-sample properties of the new methods will be established rigorously via modern empirical process theory and semiparametric efficiency theory. Efficient and stable numerical algorithms will be developed to implement the inference procedures. The proposed methods will be evaluated extensively through simulation studies mimicking real data and be applied to several major genomic studies, most of which are carried out at the UNC. Efficient, reliable and user-friendly software with proper documentation will be freely available. This research will not only advance the fields of biostatistics and statistical genetics but also influence chronic disease research at the UNC and elsewhere.