Bioconductor - DESeq

Question

[soheilazareie]

Hi We have 20 Acinar, 20 Duct, 4 control samples for RNA seq analysis, the aim is to see how Acinar are different from Duct, I have a couple of questions:

1. Is it necessary to include control samples if just the comparison of Acinar vs Duct (res_A_D <- results(ddsMF, contrast=c("condition","A","D")) is important for us? and also, how important is to have control in our DESeq analysis?

2. When I do contrast, my codes are

   res_A_WT <- results(ddsMF, contrast=c("condition","A","WT"))

   res_D_WT <- results(ddsMF, contrast=c("condition","D","WT"))

   res_A_D <- results(ddsMF, contrast=c("condition","A","D"))

My question is: why I get the same list of genes when I do the above contrast?, is it correct? if so, what is the point of doing contrast?

1. For making a heatmap of Differentially expressed genes, which one is more important, fold change or p-adjusted value?

Thanks, Soheila

[JC]

Hi Soheila, I just formatted your post and tagged as Question, because this is not a "Tutorial".

Answer 1

[JC]

Answering your 3 questions:

1. Yes, you can compare Acinar (A), Duct (D) and controls (C) but I would contrast (A / C) vs (D / C) and then get DEG between groups, that will mean to get genes differentially expressed to the control and differential expressed between A / D. But if you only care about the differences between A / D is fine just to compare both without consider C, just keep in mind that both are similar questions but not necessary are the same.

2. Are you sure the lists are the same? I would expect different genes in each class. I guess you are observing the same genes when you compare ( A/C vs D/C ) and (A/D), that can be explained if the C doesn't have an impact in the DEG.

3. Depends, if you want to show how strong are the changes, use fold-change, if you want to show how statistically significant are, use the p-value.