Complexity of gene expression data in scRNAseq

Question

[esergison]

Hello,

I'm a biologist trying to understand some scRNAseq data. I've been following this tutorial: https://github.com/hbctraining/scRNA-seq/blob/master/lessons/04_SC_quality_control.md

I'm looking at dot plots of complexity that graph unique transcripts vs sequencing reads and I'm having trouble understanding what complexity means here. Is this a way of filtering out cells with low transcript expression?

Thanks!

Answer 1

[rpolicastro]

A normal cell is expected to express (tens of) thousands of genes, so when you sequence cells with scRNA-seq you expect to capture a large number of unique genes per cell using a large number of UMIs. If you instead only capture a small number of unique genes with a large number of UMIs that usually indicates some problem. For example, RBCs have little/no RNA content so will often have only a hundred or so detected genes despite having a high number of total UMIs because you keep sequencing transcripts derived from only a few genes over and over again. It could be indicative of cell quality in general too. For example, the droplet could have captured a lysed cell that's missing most of its RNA.