CIF: Small: Collaborative Research: A General Theory of Group Testing for Genotyping

下一代高通量测序设备为人类基因组学领域带来了个性化基因分型的希望。个性化基因分型的进展取决于成本效益高的大规模关联研究程序的发展。这种大规模实验和研究面临着与生物工程系统和测试对象施加的限制有关的一些障碍，包括a)必须用个人遗传物质的小样本进行测试的限制；B)受试者必须多路复用的限制，因此基因序列必须被条形码；C)测试结果是半定量的；D)许多检测必须在相关个体的群体中进行，从而大大增加了筛查整个家庭的成本；E)测试必须应对变异发现的挑战，以及可能表现出不同基因拷贝数的个体。克服这些障碍和许多其他障碍的自然方法是通过群体测试进行基因分型。该项目的目标包括开发一个全面的，但分析或计算易于处理的基因分型组测试的一般理论。提出的理论将回答由测序引起的基因分型的独特挑战，尽管该理论的一部分可能会在不同的领域找到独立的应用，如约束多址通道分析，指纹和识别编码，以及错误控制编码。特别是，研究者在群体检验领域引入了几种新的模型，包括不同类型和优势的被试、半定量检验、二维池和泊松概率检验。

Next generation high-throughput sequencing devices have brought the promise of personalized genotyping to the field of human genomics. Advances in personalized genotyping depend on developments of procedures for cost-efficient large-scale association studies. Such large-scale experiments and studies face a number of obstacles related to constraints imposed by bioengineering systems and test subjects, including a) the restriction that tests have to be performed with small samples of individuals' genetic material; b) the constraint that subjects have to be multiplexed, so that genetic sequences have to be barcoded; c) the fact that the outcomes of the tests are semi-quantitative; d) the reality that many tests have to be performed within groups of related individuals, thereby significantly increasing the cost of screening whole families; e) the fact that tests have to deal with the challenge of variant discovery and individuals that may exhibit different gene copy numbers. A natural way to overcome these and many other obstacles is to perform genotyping via group testing.The goals of this project include developing of a comprehensive, yet analytically or computationally tractable general theory of group testing for genotyping. The proposed theory will answer the unique challenges arising in genotyping by sequencing, although parts of the theory may find independent applications in areas as diverse as constrained multiple access channel analysis, fingerprinting and identification coding, and error-control coding. In particular, the investigators introduce several new models into the field of group testing, including subjects with different types and strengths, semi-quantitative testing, two-dimensional pooling, and Poisson probabilistic testing.