Abstract:
|
Single cell RNA-sequencing (scRNA-seq) is a promising tool that facilitates study of the transcriptome at the resolution of a single cell. We previously developed SCnorm to account for the variability in the relationship between expression and sequencing depth, which we refer to as the count-depth relationship, during normalization. To investigate the source of this variability, we developed a first principles simulation framework which takes each step of generating scRNA-seq data into account. With this framework, we demonstrate the contribution of various protocol choices to technical artifacts observed in scRNA-seq data. Furthermore, we illustrate how a critical step in most scRNA-seq protocols directly contributes to the systematic variability in the count depth relationship, and show that hypotheses generated with the simulation are supported by existing independent datasets.
|