Noller Kathleen, Cahan Patrick
Institute for Cell Engineering, Johns Hopkins University, Baltimore MD USA.
Department of Biomedical Engineering, Johns Hopkins University, Baltimore MD USA.
bioRxiv. 2024 Jul 22:2024.07.19.604184. doi: 10.1101/2024.07.19.604184.
Methods that predict fate potential or degree of differentiation from transcriptomic data have identified rare progenitor populations and uncovered developmental regulatory mechanisms. However, some state-of-the-art methods are too computationally burdensome for emerging large-scale data and all methods make inaccurate predictions in certain biological systems. We developed a method in R (stemFinder) that predicts single cell differentiation time based on heterogeneity in cell cycle gene expression. Our method is computationally tractable and is as good as or superior to competitors. As part of our benchmarking, we implemented four different performance metrics to assist potential users in selecting the tool that is most apt for their application. Finally, we explore the relationship between differentiation time and cell fate potential by analyzing a lineage tracing dataset with clonally labelled hematopoietic cells, revealing that metrics of differentiation time are correlated with the number of downstream lineages.
从转录组数据预测细胞命运潜能或分化程度的方法已经识别出罕见的祖细胞群体，并揭示了发育调控机制。然而，一些最先进的方法对于新出现的大规模数据来说计算负担过重，而且所有方法在某些生物系统中都会做出不准确的预测。我们在R语言中开发了一种方法（stemFinder），该方法基于细胞周期基因表达的异质性来预测单细胞分化时间。我们的方法在计算上易于处理，并且与竞争对手一样好或更胜一筹。作为基准测试的一部分，我们实施了四种不同的性能指标，以帮助潜在用户选择最适合其应用的工具。最后，我们通过分析一个带有克隆标记造血细胞的谱系追踪数据集，探索了分化时间与细胞命运潜能之间的关系，结果表明分化时间的指标与下游谱系的数量相关。
