Australian Institute for Bioengineering and Nanotechnology, The University of Queensland, Brisbane, QLD, Australia.
Department of Genetics, Albert Einstein College of Medicine, Bronx, NY, 10461, USA.
Genome Biol. 2023 Oct 20;24(1):238. doi: 10.1186/s13059-023-03036-2.
Single-cell RNA-sequencing (scRNA-seq) technologies enable the capture of gene expression heterogeneity and consequently facilitate the study of cell-to-cell variability at the cell type level. Although different methods have been proposed to quantify cell-to-cell variability, it is unclear what the optimal statistical approach is, especially in light of challenging data structures that are unique to scRNA-seq data like zero inflation.
We systematically evaluate the performance of 14 different variability metrics that are commonly applied to transcriptomic data for measuring cell-to-cell variability. Leveraging simulations and real datasets, we benchmark the metric performance based on data-specific features, sparsity and sequencing platform, biological properties, and the ability to recapitulate true levels of biological variability based on known gene sets. Next, we use scran, the metric with the strongest all-round performance, to investigate changes in cell-to-cell variability that occur during B cell differentiation and the aging processes. The analysis of primary cell types from hematopoietic stem cells (HSCs) and B lymphopoiesis reveals unique gene signatures with consistent patterns of variable and stable expression profiles during B cell differentiation which highlights the significance of these methods. Identifying differentially variable genes between young and old cells elucidates the regulatory changes that may be overlooked by solely focusing on mean expression changes and we investigate this in the context of regulatory networks.
We highlight the importance of capturing cell-to-cell gene expression variability in a complex biological process like differentiation and aging and emphasize the value of these findings at the level of individual cell types.
单细胞 RNA 测序 (scRNA-seq) 技术能够捕获基因表达的异质性，从而促进细胞类型水平上的细胞间变异性研究。尽管已经提出了不同的方法来量化细胞间的变异性，但尚不清楚最佳的统计方法是什么，尤其是在考虑到 scRNA-seq 数据特有的挑战性数据结构（如零膨胀）时。
我们系统地评估了 14 种常用的转录组数据分析变异度量的性能，用于测量细胞间的变异性。利用模拟和真实数据集，我们根据数据特有的特征、稀疏性和测序平台、生物学特性以及基于已知基因集重现真实水平的生物学变异性的能力，对指标性能进行基准测试。接下来，我们使用表现最强的综合性能的 scran 来研究 B 细胞分化和衰老过程中细胞间变异性的变化。对造血干细胞 (HSCs) 和 B 淋巴样发生中的原代细胞类型的分析揭示了独特的基因特征，这些特征在 B 细胞分化过程中表现出一致的可变和稳定表达模式，突出了这些方法的重要性。识别年轻和年老细胞之间的差异可变基因阐明了仅关注平均表达变化可能会忽略的调控变化，我们在调控网络的背景下对此进行了研究。
我们强调了在分化和衰老等复杂生物学过程中捕获细胞间基因表达变异性的重要性，并强调了这些发现对于个体细胞类型的价值。
