CAREER: Learning Mechanisms from Single Cell Multi-Omics Data

该奖项全部或部分由《2021年美国救援计划法案》（公法117-2）资助。细胞是生命的基本单位。了解细胞如何分化成各种细胞类型以及细胞在进化过程中如何变化一直是一个长期存在的科学问题。最近的测序和成像技术可以提供组织中每个单细胞的大规模数据，包括一个基因的mRNA产物的数量，某些基因的蛋白质产物的数量，染色质三维结构，以及每个细胞的空间位置。该项目将开发计算方法，从复杂的高维数据中学习细胞分化和发育的机制。该项目将为社区提供开源工具，并可与教育活动和推广活动相结合，例如计算生物学中的算法主题课程。为高中生和本科生，特别是那些来自代表性不足群体的学生提供参与计算生物学前沿研究的机会。近年来，单细胞技术在“形态”和“规模”两方面都取得了进步。在模式方面，多组学技术允许研究人员从多个方面分析每个单细胞，包括转录组、基因组、染色质可及性和蛋白质丰度。在规模方面，更多的细胞、组织、个体和物种被单细胞rna测序（scRNA-seq）技术所描述。计算集成方法已经发展到集成来自不同模式或不同批次的单细胞数据。该项目的目标是开发以机制为重点的集成工具，并从大规模、多模态的单细胞数据中学习分子机制。具体目标是：(1)开发整合来自同一组织的成对和非成对单细胞多组学数据的方法，特别是提出一种考虑跨模态关系的共识细胞身份学习方法；(2)发展深度学习方法，整合来自多个个体或物种的scRNA-seq数据。该方法旨在消除技术批次效应，但保留数据矩阵之间的生物差异，并推断与年龄和种族等每个元特征相关的基因；(3)在时间或空间背景下研究细胞特异性基因调控网络（grn）。当细胞的空间位置可用时，该项目将实现一种方法来学习细胞特异性grn和细胞间相互作用。这项研究有可能在利用大规模、多模态单细胞数据理解细胞和疾病的机制方面向前迈出重要一步。该项目的结果可以在PI的网站上找到：https://xiuweizhang.wordpress.com/.This该奖项反映了美国国家科学基金会的法定使命，并通过基金会的智力价值和更广泛的影响审查标准进行评估，认为值得支持。

This award is funded in whole or in part under the American Rescue Plan Act of 2021 (Public Law 117-2).Cells are the fundamental units of life. Understanding how cells differentiate into various cell types and how cells change during evolutionary process has been a long-standing scientific problem. Recent sequencing and imaging technologies can provide large scale data for each single cell in a tissue, including the amount of mRNA products of a gene, the amount of protein products of certain genes, the chromatin 3D structure, and the spatial location of each cell. The project will develop computational methods to learn mechanisms in cell differentiation and development from complex and high dimensional data. The project will provide open-source tools for the community and can be integrated with educational activities and outreach, such as courses on topics of algorithms in computational biology. Opportunities for high school and undergraduate students especially those from underrepresented groups will be provided to participate in cutting-edge research in computational biology.During recent years, single cell technologies have progressed in terms of both “modality” and “scale”. In terms of modality, multi-omic technologies allow researchers to profile each single cell from multiple aspects, including transcriptome, genome, chromatin accessibility and protein abundance. In terms of scale, more cells, tissues, individuals, and species are profiled with single cell RNA-sequencing (scRNA-seq) technology. Computational integration methods have been developed to integrate single cell data from different modalities or different batches. The goal of the project is to develop integration tools with a strong emphasis on mechanisms, and to learn molecular mechanisms from the large-scale, multi-modality single cell data. The specific aims are: (1) develop methods to integrate paired and unpaired single cell multi-omics data from the same tissue, and in particular, propose a method to learn consensus cell identity considering cross-modality relationships; (2) develop deep learning methods to integrate scRNA-seq data from multiple individuals or species. The method aims to remove technical batch effects but preserve biological variation between data matrices and infer the genes which are associated with each meta feature like age and race; (3) develop methods to learn cell-specific gene regulatory networks (GRNs) for cells in a temporal or spatial context. When spatial locations of cells are available, the project will implement a method to learn both cell-specific GRNs and cell-cell interactions. This research has the potential to make a significant step forward in understanding mechanisms in cells and diseases using large-scale, multi-modality single cell data. The results of the project can be found at the PI’s website: https://xiuweizhang.wordpress.com/.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.