A scalable integrated multi-modal single cell analysis framework for gene regulatory and cell-cell interaction networks

用于高通量生物分子测量的单细胞技术的进步正在给生物学和医学的研究带来革命性的变化。虽然已经开发了用于单细胞分析中出现的各种任务的计算工具，但由于计算机内存和运行时间的限制，它们往往缺乏处理数百个数据集和/或数百万个细胞的能力。该项目解决了由于公共储存库中迅速积累单细胞数据集而产生的开发用于大规模数据分析的计算方法和软件的迫切需要，以及为了充分揭示生物系统的复杂性而对其进行分析的需要。该项目将导致对多种类型的单细胞数据进行综合分析的新方法，利用这些数据建立生物网络和了解细胞间通信，并开发多种软件产品以供更广泛地采用。项目团队将包括在跨学科研究中获得宝贵经验的学生。根据该项目开发的软件工具的使用将在亚特兰大单细胞基因组和分析倡议(https://ascomai.org/)，)下的培训讲习班上教授，该倡议为佐治亚理工学院、埃默里大学和莫尔豪斯医学院的单细胞研究社区提供服务。该项目将包括代表人数不足的个人的大量参与。该项目将导致开发可扩展的、内存高效的算法和高性能的并行实施，以利用现在常见的多插槽、多核心服务器/工作站中固有的并行性进行大规模单细胞分析。针对的问题包括1)单模式和多模式整合，2)来自不同来源的单细胞RNA测序和单细胞ATAC测序数据的聚类，3)从大规模整合的多模式单细胞数据构建基因调控网络，4)细胞内基因调控网络和细胞间相互作用网络的联合推断。这项研究将使用几个案例研究、模拟和真实世界的基准数据以及已知的黄金标准基准进行验证。研究产品将作为独立的软件工具提供，这些工具将能够从笔记本电脑到工作站再到高端共享内存服务器无缝运行，有效地利用所有可用资源来推动可分析的数据集的规模。该项目的结果将在https://faculty.cc.gatech.edu/~saluru/single-cell上公布，软件产品将在https://github.com/AluruLab.This上以开源形式发布，该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

Advances in single cell technologies for high-throughput measurement of biological molecules is revolutionizing the study of biology and medicine. While computational tools for various tasks arising in single cell analysis have been developed, they often lack the ability to handle hundreds of datasets and/or millions of cells due to both computer memory and run-time constraints. This project addresses the compelling need for the development of computational methods and software for large scale data analysis, arising due to the rapid accumulation of single cell datasets in public repositories, and the need for analyzing them to fully unravel the complexity of biological systems. The project will lead to novel methods for integrated analysis of multiple types of single cell data, using such data for building biological networks and understanding inter-cellular communication, and development of multiple software products for broader adoption. The project team will include students who will gain valuable experience in interdisciplinary research. Use of software tools developed under the project will be taught at training workshops under the Atlanta Single Cell Omics and Analytics Initiative (https://ascomai.org/), which serves the single cell research communities of Georgia Tech, Emory University, and the Morehouse School of Medicine. The project will include significant involvement of underrepresented individuals. The project will lead to the development of scalable, memory-efficient algorithms and high-performance parallel implementations for large-scale single cell analysis leveraging the inherent parallelism in a multi-socket, multi-core server/workstation that is now commonplace. The problems targeted include 1) single- and multi-modal integration, 2) clustering of single cell RNA-sequencing and single cell ATAC-sequencing data from disparate sources, 3) construction of gene regulatory networks from large-scale integrated multi-modal single cell data, and 4) joint inference of intra-cell gene regulatory networks and cell-cell interaction networks. The research will be validated using several case studies, simulated and real-world benchmark data, and known gold standard benchmarks. The research products will be made available as standalone software tools that will be able to run seamlessly from laptops to workstations to high-end shared memory servers, efficiently exploiting all available resources to push the scale of datasets that can be analyzed. Results of the project will be made available at https://faculty.cc.gatech.edu/~saluru/single-cell and software products will be released as open source on Github at https://github.com/AluruLab.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.