Design and Analysis of Displayed Peptidomes

项目总结 免疫系统直接或间接地参与了人类健康和疾病的许多方面。 然而，准确确定人类免疫反应的特定分子靶点的方法有 缺乏。我们率先使用了噬菌体免疫沉淀测序(‘PhIP-Seq’)，这是一种大规模的 涉及噬菌体展示多肽文库的多重抗体分析技术。这些 多肽由长的、高质量的合成DNA寡核苷酸文库编码。PhIP-Seq协议分析 实验使用高通量DNA测序。该技术的优点，包括样品 吞吐量和单位样品成本，使PhIP-Seq成为推动未来发展的不可或缺的工具 生物医学发现。 可以使用合成DNA编码的文库类型受到我们当前设计方法的限制。 例如，我们已经将人类蛋白质组和人类病毒体编码为~250K和~100K的多肽 库。这些文库可用于研究自身抗体反应或病毒感染的作用 例如，在复杂的疾病中。然而，由于以下原因，更大的蛋白质文库无法进行编码 成本约束。该项目的目标1致力于创新的基于k-mer的设计策略，该策略将使 表示更复杂的蛋白质空间，例如人类肠道微生物区系的集体蛋白质组。 PhIP-Seq生成一种独特的数据类型，无法使用以前开发的或 重新调整用途的软件。在这个项目的目标2中，我们试图开发基于现代的方法和软件 统计抽样理论中的方法，包括经验和完全贝叶斯方法，用于检测 抗体与多肽结合的相互作用。此外，我们建议开发一套关键的实验 有助于确保与PhIP-Seq研究相关的研究结果具有可重复性的注释标准。 最常用的PhIP-Seq试验设计涉及纵向和/或分组设计 比较。在目标3中，我们建议开发开源的BioConductor和‘Siny App’软件包， 实现典型的分析管道，以供非程序员调整以适应其特定的 做实验。这些管道将提供表位水平的分析，并重要的是考虑抗体交叉 相似蛋白质序列之间的反应性。将进行三项PhIP-Seq研究来说明新的 设计和分析软件工具：1型糖尿病研究、炎症性肠病研究和一项研究 阿尔茨海默氏症。这些产生的数据将提供给社区进行重新分析和数据 探险。

PROJECT SUMMARY The immune system is either directly or indirectly involved in many aspects of human health and disease. However, methods to accurately determine the specific molecular targets of human immune responses are lacking. We have pioneered the use of Phage ImmunoPrecipitation Sequencing (‘PhIP-Seq’), which is a massively multiplexed antibody profiling technology involving libraries of bacteriophage-displayed peptides. These peptides are encoded by long, high quality synthetic DNA oligonucleotide libraries. Analysis of PhIP-Seq experiments uses high throughput DNA sequencing. Favorable features of the technology, including sample throughput and per sample cost, uniquely position PhIP-Seq to become an indispensable tool for driving future biomedical discoveries. The types of libraries that can be encoded using synthetic DNA are limited by our current design approach. For example, we have encoded the human proteome and the human virome as ~250K and ~100K peptide libraries, respectively. These libraries can be used to study autoantibody responses or the role of viral infection in complex diseases, for example. Much larger libraries of proteins, however, are inaccessible to encoding due to cost constraints. Aim 1 of this project is devoted to an innovative ‘k-mer’ based design strategy that will enable representation of more complex protein spaces, such as the collective proteome of the human gut microbiota. PhIP-Seq produces a unique type of data, which cannot be properly analyzed using previously developed or repurposed software. In Aim 2 of this project, we seek to develop methods and software based on modern approaches in statistical sampling theory, including Empirical and Fully Bayesian approaches, for the detection of antibody-peptide binding interactions. In addition, we propose to develop a critical set of experimental annotation standards that will help to ensure that findings associated with PhIP-Seq studies are reproducible. The most commonly employed PhIP-Seq experimental designs involve longitudinal and/or group-wise comparisons. In Aim 3, we propose to develop open source Bioconductor and ‘Shiny App’ software packages that implement typical analytical pipelines for adaptation by non-programmers to the analysis of their specific experiment. These pipelines will provide epitope-level analyses, and importantly consider antibody cross- reactivity among similar protein sequences. Three PhIP-Seq studies will be performed to illustrate the new design and analysis software tools: a study of type 1 diabetes, a study of inflammatory bowel disease, and a study of Alzheimer’s disease. These resulting data will be made available to the community for re-analysis and data exploration.