How to get fraction of unspliced reads for specific gene from scRNAseq Cell Ranger output

Question

[codezero]

Hi!

I'm new to this forum (and to bioinformatics in general), so pardon my ignorance. I looked around and couldn't find a similar question.

I have the full Cell Ranger output (including BAM and .gtf files) for some scRNA-seq experiments and would like to get a column that represents what percentage of reads from a specific gene align to spliced vs unspliced regions for each barcode.

Is there an easy way to do this? If it's possible in python even better, since that's what I am mostly familiar with in terms of programming languages!

I love the forum by the way, as well as the handbook!

Thanks in advance!

Answer 1

[Rob]

If you have the raw data, you can easily get this information using alevin-fry. It processes the data such that the output contains counts for each gene, separated by whether the UMI was assigned to a spliced or unspliced molecule (or was ambiguous in splicing origin).

Answer 2

[benformatics]

I don't think you can get that information specifically: https://kb.10xgenomics.com/hc/en-us/articles/360021902671-Can-we-detect-splice-variants-or-isoforms-using-single-cell-3-RNA-seq-data-

However you could get the counts including the introns and potentially compare if you want a gene-level comparison: https://support.10xgenomics.com/docs/intron-mode-rec

You would probably need to run a different type of tool to get exactly what you want - i would try a velocity based approach: http://velocyto.org/velocyto.py/tutorial/cli.html#preparation