Establishment of Active Chromatin Domains

活性染色质结构域的建立

• 批准号: 10410617
• 负责人: Erica Nicole Larschan
• 金额: $ 7.42万
• 依托单位: BROWN UNIVERSITY
• 依托单位国家: 美国
• 财政年份: 2018
• 资助国家: 美国
• 起止时间: 2018-04-01 至 2023-03-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/10410617
• 关键词: 3-Dimensional; Biological; Cells; ChIP-seq; Chromatin; Communities; Computer Models; Data Analyses; Data Set; Development; Disease; Drosophila genus; Embryo; Funding; Gene Activation; Gene Expression Regulation; Goals; Graph; Hi-C; Joints; Machine Learning; Measurement; Methodology; Methods; Modeling; Process; Publications; Research; Students; Technology; Time; base; data modeling; graduate student; innovation; insight; mRNA sequencing; neural network; single-cell RNA sequencing; spatiotemporal; tool

RESEARCH SUMMARY: Understanding the mechanisms which drive coordinate gene activation will provide critical biological insights into development and disease. This supplement aims to enable AI/ML methods to integrate diverse datasets to help achieve that goal. Specifically, we plan to computationally model the spatiotemporal regulation of genes in Drosophila by combining ChIP-Seq, Hi-C/ Micro-C, and mRNA-Seq datasets. Moreover, we will obtaining and model single-cell measurements (scRNA-Seq and scATAC-Seq) across early embryonic developmental stages when domains of coordinate gene activation are established. Therefore, we propose three tasks to produce AI/ML-ready datasets for the community. (1) Development of robust preprocessing pipelines for easy access and input to graph-based neural networks for integrating datasets. (2) Application of sub-graph selection methods to extract informative interactions from large and noisy graphs generated from chromatin capture technologies. (3) Integrating single-cell measurements and providing these combinations for data modeling and analysis purposes. The proposed methods will assist the goals of accurately capturing the underlying biological information when using various machine learning approaches. Furthermore, we will provide processed datasets and scripts to the community to enable methodological innovations. Dr. Larschan will collaborate with machine learning expert Dr. Singh and we have collaborated on several projects based on the proposed topic with joint students and publications. This supplement will be essential to allowing us to continue collaborating and generating new methods and data sets which will provide key insight into coordinate gene activation.

研究总结： 了解驱动协调基因激活的机制将提供关键的 对发育和疾病的生物学见解。该补充旨在实现AI/ML 方法来整合不同的数据集，以帮助实现这一目标。具体来说，我们计划 在果蝇基因的时空调控的计算模型， ChIP-Seq、Hi-C/ Micro-C和mRNA-Seq数据集。此外，我们将获得和模型 早期胚胎发育过程中的单细胞测量（scRNA-Seq和scATAC-Seq） 协调基因激活域建立的阶段。因此，我们建议三个 为社区制作AI/ML就绪数据集。(1)发展强有力的 预处理管道，便于访问和输入到基于图形的神经网络， 数据集。(2)应用子图选择方法提取信息交互 由染色质捕获技术生成的大而嘈杂的图。(3)集成单电池 测量并提供这些组合用于数据建模和分析目的。的 所提出的方法将有助于准确捕获潜在的生物学特征， 使用各种机器学习方法时的信息。此外，我们将提供 处理数据集和脚本提供给社区，以实现方法创新。博士 Larschan将与机器学习专家Singh博士合作， 与联合学生和出版物共同开展了几个基于拟议主题的项目。这 补充将是必不可少的，使我们能够继续合作，并产生新的 方法和数据集，这将提供关键的洞察协调基因激活。