Entering edit mode
ti243
•
0
@ti243-9668
Last seen 9.1 years ago
Hi All,
I have a 600 single cell RNA-seq samples. And I would like to do DeSeq between 300 vs 300.
When I use the standard code, it runs for hours. And doesn't finish.
I have tried specifying 4 course but it still runs very long! EdgeR or DeSeq1 in comparison runs in a minute.
When it estimates gene-wise dispersions it becomes very slow.
Any ideas how I can make it faster for large datasets (>500)?
Thanks,
Tomi
Additionally, I doubt that the negative binomial statistical model makes sense for single cell data. There have been a flurry of papers recently showing that new statistical models are necessary to model the abundance of zeros for example.