Leveraging deep sequencing data to understand antibody maturation

描述（由申请人提供）：智力优点：每个人的健康都严重依赖于其特定的免疫系统。免疫系统的适应性成分是身体学会识别病原体的手段。适应性免疫的缺陷使个人和人群面临感染传染病和癌症的风险。目前可用的数学和计算工具尚未准备好表征抗体介导的适应性免疫系统在暴露于新的致病实体时发生的全部变化。特别是，最先进的方法一次只关注免疫细胞的一小部分，使用不是从数据中得出的免疫细胞成熟的简单模型，并且只对这些模型的参数给出点估计，这阻碍了它们的发展。研究人员建议通过开发一种新的方法来解决这些限制，通过开发：1)免疫细胞成熟的第一个完全贝叶斯推理方法；2)第一个抗体细胞成熟和进化的综合统计模型，包括直接从数据推断的抗体体细胞超突变序列模型；3)创新的推理工具，以获得项目集合到离散参数的联合分配的后验分布-这些模型和推理框架的可扩展计算实现导致它们的广泛应用。简而言之，我们的工作将为最近开发的数据类型开发急需的分析方法，并打开统计研究的新领域。更广泛的影响：免疫细胞受体高通量测序的综合统计建模和推断将为合理的疫苗设计、感染易感性预测和理解免疫细胞癌的发病机制提供所需的信息。B细胞谱系重建将使科学家能够追踪抗体对病原体进化的反应所发生的变化，使疫苗能够比病原体领先一步。这一方法的延伸将是利用这些工具不仅确定个人的免疫力，而且确定群体的免疫力，例如他们抵抗流行病的能力。我们的形式化将激发对具有挑战性的统计方面的新型推理问题的研究。我们的方法将在开源软件中实现，因此任何免疫学实验室或诊所都可以使用这些新方法。此外，提出的统计方法应该找到免疫学以外的其他应用，例如宏基因组学。

DESCRIPTION (provided by applicant): Intellectual Merit: The health of each human being is critically dependent on its particular immune system. The adaptive component of the immune system is the means by which the body learns to recognize pathogens. Deficiencies in adaptive immunity place the individual as well as the population at risk for infectious diseases and cancers. The currently available mathematical and computational tools are not yet ready to characterize the full collection of changes in the antibody-mediated adaptive immune system occurring in response to exposure to new pathogenic entities. In particular, state-of-the-art methods are hindered by only focusing on a small subset of the immune cells at a time, using simple models of immune cell maturation that are not derived from data, and only giving point estimates for parameters of those models. The investigators propose to address these limitations by developing a novel approach to high throughput sequencing data from antibody genes by developing: 1) the first fully Bayesian inferential approach to immune cell maturation; 2) the first comprehensive statistical model of antibody cell maturation and evolution, including sequence models of antibody somatic hypermutation inferred directly from data; 3) innovative inferential tools to obtain posterior distributions on the joint assignment of collections of itemsto discrete parameters - scalable computational implementations of these models and inferential frameworks leading to their widespread application. In short, our work will both develop much needed analytical methods for a recently developed type of data and open a new area of statistical research. Broader Impacts: Comprehensive statistical modeling and inference of high throughput sequencing of immune cell receptors will provide information needed for rational vaccine design, prediction of susceptibility to infections, and understanding of the pathogenesis of immune cell cancers. B cell lineage reconstructions will allow scientists to track the changes that happen to an antibody in response to pathogen evolution, enabling vaccines to stay one step ahead of pathogens. An extension of this approach will be to use these tools to characterize not only the immunity of individuals, but also of populations, for example in their ability to resist epidemics. Our formalization will motivate research on a new type of inference problem with challenging statistical aspects. Our methods will be implemented in open-source software, so that any immunology lab or clinic can use these new approaches. Moreover, the proposed statistical methodology should find other applications beyond immunology, for example, in metagenomics.