Entering edit mode
Seaweed
•
0
@seaweed
Last seen 3.7 years ago
Hi,
I have run a bulk RNA-seq for 4 tissue samples. Each sample contains 50 cells collected from FACS, but due to technical challenges, my cells of interest account for varying proportions of those 50 cells across the 4 preps. Are there any functions in DESeq2 or other packages to uncover genes that are expressed in my cells of interest even though the samples were a mixture of cell types? My cells are labeled by tdTomato, so I do have counts for tdTomato transcripts. One thought is to find genes that would correlate with tdTomato expression. Is this reasonable? If so, what would be a good way to look for correlated genes in the whole dataset? Clustered heatmap seems to work best for selected set of genes?
Thanks!
Unfortunately, no. It is a cell type that not many have worked on. But I do know that one of my samples contain none of my cells of interest.
We have a function
unmixwhich performs estimation of composition whenpureexpression profiles are available, using the VST implemented in DESeq2, but it wouldn't work in this case.I think if you have a single expression profile, a method from Wenyi Wang's group can be used:
https://bioinformatics.mdanderson.org/public-software/demixt/
Thank you for the suggestions! It is actually our goal to get the single expression profile of our cells, but this is technically challenging as we are working with tissues. Is it not recommended to look for genes that correlate with our markers to at least get a sense of what candidate genes may be expressed in our cells? Thanks!
I guess that sounds fine but I don't have any particular recommendations or DESeq2 functions to help you there.
No problem. Thanks!