Understanding how to deal with Brain snRNA-seq data

Question

[AlexStar]

I have obtained normal brain snRNA-seq data, but due to my limited experience and knowledge in this tissue, I need guidance on applying pre-processing steps.

Firstly, what is the usual threshold for filtering out mitochondrial genes based on their percentage of total counts? In one brain tumor dataset, a 5% cutoff was used (i.e., cells with more than 5% mitochondrial gene counts were removed). Is this the standard for this tissue?

Additionally, what is the cutoff for the number of genes with at least one count in a cell (in Scanpy, this is n_genes_by_counts)?

These cutoffs can significantly impact the analysis, so it's crucial for me to get them right.

Note: The data at hand is composed of neurons and glial cells. Should the thresholds change for each cell type?

Answer 1

[ATpoint]

Please don't ask the same question twice, it was answered by multiple people before: snRNA-seq of the healthy human brain

Note: The data at hand is composed of neurons and glial cells. Should the thresholds change for each cell type?

Maybe yes, I often put celltyper-specific thresholds. For example in immune cells, monocyte express many thousands of genes in scRNA-seq while neutrophile express ~ 1500 in most conditions. Cutoffs are different here.

[AlexStar]

ATpoint All I need is the mitochondrial cutoff used in the brain. I have no idea how to decide that. I hoped someone here has knowledge in this domain and could help.

[Ming Tommy Tang]

Take a look at this paper: https://pubmed.ncbi.nlm.nih.gov/32840568/