single cell RNA seq Differential Expression using MAST

Question

[quintanamaster]

Hi all,

I'm using MAST package to detect differentially expressed genes in single cell RNAseq datasets. I've used DESeq2 in the past and I know that it selects the reference group for calculating fold changes alphabetically if not pre-defined. I'm struggling to understand if MAST does the same thing? Does it select the reference group for FC calculation alphabetically? It's not very clear from the documentation available.

Thanks in advance for helping,

Juan

Answer 1

[ATpoint]

I guess so. The keyword you are looking for is reference level, see https://www.bioconductor.org/packages/release/bioc/vignettes/MAST/inst/doc/MAITAnalysis.html#4_differential_expression_using_a_hurdle_model

The relevel function is what you need as described in the first few lines of code of paragraph 4.

[quintanamaster]

Thanks ATpoint!

I'm running the code from the Seurat vignette as below:

head<- FindMarkers(Combined, ident.1 = "Ast_2", ident.2 = NULL, test.use = "MAST") View(head) write.csv(head, "../Desktop/Hyp_DE_Ast22.csv", row.names = TRUE)

It doesn't let me use the "relevel" option so that's why I was wondering if it does the selection alphabetically by default?

[igor]

It looks like you are actually using Seurat, not MAST. Although you are using the MAST test, you are using Seurat's interface, not the native MAST one. Thus, you have to refer to Seurat documentation.

If you are not sure what the order is, check the average expression (AverageExpression()) and then you can see which group is higher or lower.