Suggestions for doing differential expression with 10x single cell data (two conditions, 4 replicates each, having 30K cells in each conditions))

Question

[m.sadman.sakib]

Hello everyone!

I have just started analysing a 10x single cell dataset, which has two conditions(control vs treatment). I am using Seurat and their detailed vignettes are of greatly helpful! Along with other exploratory analysis that is beautifully demonstrated there, I want to perform the differential expression analysis, within different cell types(for example, Cell A, control vs treatment, then Cell B, C,....up to 15 different cell types). To point out, all of the 8 samples were processed and sequenced at the same time, so no batch effects are there. Now, my exploratory readings have brought me down to these different versions of analysis pipelines.

1. Using just Seurat, but this is not directly using any dedicated diff. expression packages like deseq2 or edgeR Tutorial: Integrating stimulated vs. control PBMC datasets to learn cell-type specific responses

2. Using Seurat::FindMarkers, using many different statistical methods, including DESeq2. For example, FindMarkers(Object, ident.1 = "Cell type A", ident.2 = NULL, only.pos = TRUE, test.use = "DESeq2"). But the problem is, by ident.2 = NULL, this will compare "Cell type A" with all the other cell types, not within that particular cell type. Also, I am not sure how I can implement this in this case, but just putting it here for reference.

3. From OSCA book Orchestrating Single-Cell Analysis with Bioconductor. There is a whole chapter dedicating to multi-sample, pseudobulk diff. expression analysis in pairwise comparisons, within individual cell types. It is using primarily EdgeR, along with many other useful packages that simplify my problem. I found it while compiling this question here in biostars, haha!

4. DESeq2 vignette has a small paragraph giving recommendations about analyzing DESeq2 Vignette::Recommendations for single-cell analysis.

5. Also a useful resource by Harvard Chan Bioinformatics Core training on Github, explaining pseudobulk-deseq2 analysis framework Single-cell RNA-seq: Pseudobulk differential expression analysis. Seems very useful!

Now, as I am just starting to analyze, I would really appreciate, if any of you could comment on what is the best and common practice for performing diff. expression analysis for 10x single cell RNAseq. Of course I can try all 5 of the above-mentioned procedures, but what is the best and common practice in the community? Your comments and suggestions will be extremely helpful for me and will save me a lot of time! Thank you in advance for your help!

I would like to take this opportunity to thank this awesome and amazing community and its wonderful people! Thank you from the bottom of my heart!