search

broadinstitute / CellBender

CellBender is a software package for eliminating technical artifacts from high-throughput single-cell RNA sequencing (scRNA-seq) data.
https://cellbender.rtfd.io
BSD 3-Clause "New" or "Revised" License
279 stars 50 forks source link

Cell Bender + RNA Velocity #108

Open samuel-marsh opened 3 years ago

samuel-marsh commented 3 years ago

Hi Stephen & Team,

Another "quick" question for you. I'm wondering if you have any thoughts on use of Cell Bender with RNA Velocity based analyses?

One thought I had was to basically generate vectors of pre>post CB data for each gene in each sample using the spliced counts from cell ranger and then used same vector to "remove background" from unspliced counts (from velocyto). However, that assumes that background is equivalent between spliced and unspliced which I have no idea whether that is actually the case.

Alternatively could create a pseudo-10X H5 file from unspliced count matrix (Using DropletUtils::write10xCounts) and just run Cell Bender separately on spliced and unspliced counts and then recombine the outputs.

Any other thoughts you have on the issue/question would be great!

Best, Sam

sjfleming commented 3 years ago

Hi @samuel-marsh , this is an interesting question. I'm going to tag in @mbabadi as well.

I think you are probably right that the "spliced background" and the "unspliced background" might not necessarily be the same, if I am thinking about this correctly.

Processing a sample in two pieces: spliced count matrix and unspliced count matrix, does seem like a possibility.

I have little hands-on experience running RNA velocity analyses myself. If you can create two separate count matrices and remove-background from them separately, then conceptually that strikes me as the "right" way to go about it.

Do you agree @mbabadi ?