This is a test version of Biostars. For the public version, visit https://www.biostars.org.
DGEA using scRNA seq data.

I performed DGEA on scRNA-seq data in two ways. First, on the whole data, and second, per cluster. For both ways, I used EdgeR and DESeq2. The whole data is not showing any significant DEGs, upregulated or downregulated in healthy or diseased conditions.enter image description here

Whereas, when done cluster wise most of the clusters are showing some significant DEGs.enter image description here Can anyone tell why is this happening or if i have done something wrong? Thanks!!

seurat dgea seq scrna
• 433 views • link
updated by nmabbasi • nmabbasi • written by yadavsoumya647 • yadavsoumya647

We need more information on your conditions, your species, your cell types... and your code.

• 0 views • link
written by Bastien Hervé Bastien Hervé

Sure! I am working with samples from three different dataset, trying to find the differences in the number of cells in healthy and diseased condition. The tissue is human nucleus pulposus of the intervertebral disc and the cell are yet to be annotated but they are expected to be the general nucleus pulposus cells. First i performed the DGEA without pseudobulking using the general seurat pipeline which is mentioned below-

np_integrated <- NormalizeData(np_integrated )

Idents(np_integrated ) <- "orig.ident"
np.markers <- FindMarkers(np_integrated , ident.1 = "Diseased", ident.2 = "Healthy")

This gave us a plot which looked like this- enter image description here

Then we performed pseudobulking using both edgeR and DESeq2 individually:-

pseudo_np <- AggregateExpression(np_integrated, assays = "RNA", return.seurat = T, group.by = c("orig.ident", "seurat_clusters"))

Idents(pseudo_np) <- "orig.ident"

bulk.mono.de <- FindMarkers(object = pseudo_np, 
                         ident.1 = "Diseased", 
                         ident.2 = "Healthy",
                         test.use = "DESeq2")

Which resulted in the plot pasted in my original question.

• 0 views • link
updated by GenoMax GenoMax • written by yadavsoumya647 • yadavsoumya647

trying to find the differences in the number of cells in healthy and diseased condition

For this you need to look for compotition analysis tools, like scCODA.

First, run your single cell analysis to get your cluster that you can annotate. Once you have located your cell type of interest you can aggregate per celltype/replicate/condition and run your DGEA

• 0 views • link
written by Bastien Hervé Bastien Hervé

Okay Thanks! Is it possible to observe significant DEGs when we perform per cluster DGEA and observe no significant DEGs when done on the data as a whole? Also is pseudobulking preferred while performing DGEA?

• 0 views • link
written by yadavsoumya647 • yadavsoumya647

In theory yes, if the clusters DEGs are balancing each others, but pretty unlikely, especially with FDR=1 for all genes. Yes, the only acceptable situation to use Wilcoxon test is when comparing 2 clusters, when comparing condition always pseudobulk.

• 0 views • link
written by Bastien Hervé Bastien Hervé

0 answers

No answers yet.

Log in to answer this question.

More posts like this