CAREER: Statistical Models and Parallel-computing Methods for Analyzing Sparse and Large Single-cell Chromatin Interaction Datasets

核染色质的三维结构在维持正常细胞功能方面起着至关重要的作用。单细胞Hi-C(Schi-C)技术的最新发展使研究人员能够描绘单个细胞中全基因组染色质的相互作用，并回答基本的生物学问题。然而，由于数据稀疏性、高度多样性和复杂的染色体组织层次结构，用于分析Schi-C数据的计算方法在很大程度上是滞后的。该项目将通过为实验和计算生物学家建立一套模型和并行工具来分析稀疏和大型Schi-C数据集来应对这些挑战，这将帮助他们了解3D染色质相互作用并深入了解功能结果。该项目产生的饱和模型将大大提高实验质量，增强数据分析能力，降低实验成本。综合研究和教育活动包括为计算机科学、统计学、生物信息学和生物学等跨学科学科的本科生和研究生课程提供多样化的学生培训和课程开发。本项目的目标是利用随机理论、小波变换和并行计算技术研究不同尺度上特定细胞的染色质组织。首先，建立随机模型，通过评估主要协议步骤和排序深度来了解Schi-C数据的饱和状态。其次，基于多尺度的方法将被设计用于检测拓扑相关的结构域和蛋白质介导环、细胞聚集和比较分析。第三，所有模型和工具将完全并行化，以处理大数据集。这些方法的高性能和运行速度将帮助研究人员一站式实现生物发现，并节省实验和计算分析的时间和金钱。软件流水线将由PYTHON/R语言实现，并通过GitHub存储库https://github.com/chenyongrowan和罗文大学的网络服务器公开访问。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。

The 3D organization of nuclear chromatin plays a critical role in maintaining normal cellular functions. Recent development of single-cell Hi-C (scHi-C) technologies allows researchers to delineate the genome-wide chromatin interactions in individual cells and answer fundamental biological questions. However, computational methods for analyzing scHi-C data are largely lagging behind because of data sparsity, high diversity and complicated hierarchy of chromosomal organization. This project will address these challenges by building a suite of models and parallel tools for both experimental and computational biologists to analyze sparse and large scHi-C datasets, which will help them understand 3D chromatin interactions and gain deep insights into functional outcomes. The saturation model produced by this project will lead to a major improvement in experimental quality, enhanced data analysis, and reduced experimental costs. The integrated research and educational activities include diverse students training and curriculum development for both undergraduate and graduate courses in interdisciplinary subjects spanning computer science, statistics, bioinformatics and biology. The objectives of this project are to study the cell-specific chromatin organization at different scales by using stochastic theory, wavelet transform, and parallel-computing techniques. First, a stochastic model will be established to understand the saturation status of scHi-C data by evaluating the major protocol steps and sequencing depth. Second, multiscale-based methods will be designed for detecting topologically associated domains and protein-mediated loops, cell clustering, and comparative analysis. Third, all models and tools will be fully parallelized for processing large datasets. The high performance and running speed of these methods will help researchers achieve biological discoveries in a one-stop service, and save time and money for experiments and computational analysis. The software pipelines will be implemented by Python/R languages and publicly accessible through the GitHub repository https://github.com/chenyongrowan and a webserver at Rowan University.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.