Inputs for CellRanger pipeline/Loupe

Question

[ww22runner]

Hi everyone,

I have a tpm gene expression matrix from a publication which I would like to create a T-SNE plot for, and get a general sense of gene expression levels for. However, I do not have the fastq files which I believe are the inputs for cellranger-atac mkfastq pipeline. Is there any way I can use my tpm matrix instead? I would ultimately like to look at this information using Loupe but I am wondering what inputs it accepts or if I somehow have to generate a cLoupe file through the cell ranger pipeline. Any advice is greatly appreciated!

Thank you!

Answer 1

[jared.andrews07]

Is this from bulk or single-cell RNA-seq data? If the latter, there's nothing keeping you from creating a Seurat or SingleCellExperiment object from your tpm counts in the data slot for each and generating a tSNE plot from that via Seurat or scater. Though I'd only use any findings from it on an exploratory basis. Likely easier than trying to munge stuff into the Loupe format.

[ww22runner]

Hi Jared,

This is for single-cell RNA seq data and thank you for your advice. May I ask what you meant by " Though I'd only use any findings from it on an exploratory basis"?

[swbarnes2]

He means that normalized counts are fine for making pictures, but to find DE genes, you need raw counts, and using something else is going to make your math conclusions suspect.

[jared.andrews07]

Exactly this. If you just want to visualize count distributions across clusters, it'll probably be fine to give you an idea of how your gene of interest is expressed, but you really need to go from the raw data for anything you plan to publish.

[ww22runner]

Thank you, this was valuable advice.

Answer 2

[swbarnes2]

I'm not familiar with the 10x ATAC pipeline, but I know even 10x matrix files are not good enough to make a Loupe file for their regular RNASeq protocol.

You will probably need fastqs files to make the Loupe output.