search

yelabucsf / scrna-parameter-estimation

Direct estimation of mean and covariance from single cell RNA seq experiments
MIT License
76 stars 6 forks source link

Analyzing designs with multiple treatments + chemistry question #26

Open LinearParadox opened 3 weeks ago

LinearParadox commented 3 weeks ago

Is there a capture rate you recommend using for GemX chemistries? (https://www.10xgenomics.com/support/single-cell-gene-expression/documentation/steps/library-prep/chromium-gem-x-single-cell-3-reagent-kits-safety-data-sheets-v4-chemistry)

Also, I wanted to ask if there's any recommendation on how to analyze designs with multiple conditions. For example:

treat1-r1: treat1 treat1-r1: treat1 treat2-r1: treat2 treat2-r2: treat2 treat3-r1: treat3 treat3-r2: treat3

Currently I'm just sub setting the anndata to the comparisons of interest, but is there a cleaner way of doing this?

Thanks all!

mincheoly commented 2 weeks ago

I would use the same capture_rate as you would for v3 chemistry kit for now: 0.25 * sequencing saturation.

I think this would depend on the nature of the conditions - are the treatments related (levels/dosage of a drug)? If they are completely discrete, I think separating the analyses makes sense.

LinearParadox commented 2 weeks ago

I would use the same capture_rate as you would for v3 chemistry kit for now: 0.25 * sequencing saturation.

I think this would depend on the nature of the conditions - are the treatments related (levels/dosage of a drug)? If they are completely discrete, I think separating the analyses makes sense.

Got it. Is there any way to test the capture rate parameter? I ask because at least the V4 experiments I've analyzed, the I'm seeing more features detected with less sequencing depth than prior chemistries, and pretty notably so.

Got it, this is what I've been doing, I just wasn't sure if there was a "cleaner" way of doing it (like contrasts in R) that I was missing.