Formula for Differential Expression Analysis of 2 conditions and 2 cell types

Question

[hkarakurt]

Hello everyone, I have a data set with 12 samples belong to 2 conditions and 2 cell types. It is a temperature experiment. My conditions are: 4 of Cold Cell A , 3 of Warm Cell A and 5 of Warm Cell B. Cell B and Cell A can transform each other so I want to include it too. Is it possible to identify the AS events coming from temperature changes but not from cell type differences?

The tool I am using is based on edgeR and I need to give a formula to the function. I prepared a formula but I am not totally sure that if it works properly:

targets <- data.frame(row.names = paste0('Sample',c(1:12)),
                  bam = BAMFiles[seq(3,25,by=2)],
                  temperature = c(rep("Cold",4),rep("Warm",8)),
                  cell_type = c(rep("Beige",7) , rep("White",5)),
                  stringsAsFactors = FALSE)

form <- formula(~0 + temperature+cell_type+temperature:cell_type)
model.matrix(form,targets)

Do you think this formula is okay or do I need to change it?

Thank you in advance

Answer 1

[Gordon Smyth]

We always recommend that you code up a single factor to distinguish your four experimental groups (Cold.A, Warm.A, Cold.B, Warm.B) because factorial models are complex and do not do what you think they do. See the edgeR or limma User Guide or here:

https://bioconductor.org/packages/release/workflows/vignettes/RNAseq123/inst/doc/designmatrices.html

If you take the single-factor approach, then testing the hypothesis you want is quite easy. With the factorial model that you have now, testing your hypothesis is nearly impossible.

What is an "AS event"? Alternative splicing perhaps? Not sure how you're planning to explore that. My answer would apply for either differential expression or for diffSpliceDGE.

[hkarakurt]

Yes AS events is alternative splicing events. I am using ASpli package for this analyses but this package uses edgeR so I thought a generalized question would be okay.