Collaborative Research: Statistical Optimization for Barcoding and Decoding Single-Cell Dynamics via CRISPR Gene Editing

基因编辑和单细胞测序等新的基因组学技术使人类细胞结构和功能的定义取得了巨大进展，其分辨率和规模前所未有。传统的“批量”基因组测量通常是不够的，因为样品异质性和差异细胞状态动力学可能被掩盖。基于crispr的基因编辑技术可以使用可进化的条形码对单细胞谱系进行跟踪。在此技术的基础上，我们将通过开发在线优化和单细胞时间序列分析方法来解决单细胞条形码和解码所带来的数学挑战。这条研究路线将产生条形码单细胞数据的分析工具，具有可证明的理论保证，既可共享又可部署，用于定义细胞谱系和基因表达状态的遗传基础。这些工具的应用将有助于揭示细胞进化和健康与疾病相互作用背后的复杂生物学。这项研究包括适合学生参与的项目和不同层次的培训，以及开源软件的开发。目前的蜂窝条形码工具受限于其可扩展性，并且条形码技术使单细胞转换数据成为可能，这对分析提出了新的挑战。为了显著提高细胞条形码的容量，我们将开发学习理论优化方法，以便在大的组合设计空间中有效地找到最佳条形码设计。此外，我们将开发状态嵌入方法，从基因表达路径中识别细胞状态的数学抽象，由条形码轨迹支持，这意味着关键的细胞动力学和遗传调控。特别是，我们将使用核化低秩近似和凸多肽近似方案来估计亚稳态细胞簇和基因表达标志。最后，我们将使用真实癌症模型实验的数据验证动态解码工具。这一研究成果将为生物驱动的数学方法开辟新的潜力。它们将对干细胞和发育生物学、神经科学、免疫学和其他生物研究领域产生影响。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知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Novel genomics technologies, such as gene editing and single-cell sequencing, have brought great progress on defining the structure and function of human cells at unprecedented resolution and scale. Traditional “bulk” genomic measurements are often insufficient as sample heterogeneity and differential cell state dynamics could be masked. CRISPR-based gene-editing technology enables single-cell lineage tracking using evolvable barcodes. Building on this technology, we will address mathematical challenges arising from single-cell barcoding and decoding, by developing online optimization and single-cell time series analysis methods. This line of research would generate analytical tools for barcoded singlecell data, with provable theoretical guarantees, that are both sharable and deployable for defining the genetic basis of cellular lineage and gene expression states. Application of these tools will help uncover the complex biology behind cell evolution and interaction in health and diseases. The research includes projects suitable for student participation and training at various levels, and open-source software development. Current cellular barcoding tool is limited by its scalability, and the single-cell transition data made possible by the barcoding technology presents new analytical challenges. To significantly improve the capacity of cell barcoding, we will develop learning-theoretic optimization methods for efficiently finding the best barcoding design over a large combinatorial design space. Further, we will develop state embedding methods to identify mathematical abstractions of cell states from gene-expression paths, as supported by the barcoded trajectories, that imply critical cell dynamics and genetic regulation. In particular, we will use kernelized low-rank approximation and convex polytope approximation schemes to estimate metastable cell clusters and gene-expression landmarks. Finally, we will validate the dynamics decoding tools using data from real cancer model experiments. The research findings will open up new potential in biological-driven mathematical methods. They will have implications for stem-cell and developmental biology, neuroscience, immunology, and other areas of biological investigation.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.