This is a test version of Biostars. For the public version, visit https://www.biostars.org.
DESeq2 Multifactor Design

I am confused about multiple factor design and contrasting following differential expression analysis. I've been reading posts about this topic (e.g., DESeq2 design with multiple conditions ), but I wanted some clarification.

My experimental design is as follows: Two breeds of cows (HA and CA). Within each breed, there is an infected and uninfected group. The comparisons I want are: HA_infected vs. HA_uninfected, CA_infected vs. CA_uninfected, and HA_infected vs. CA_infected. Right now, my DESeq2 design is design =~ breed + status + breed:status. First of all, is this correct?

Also, how do I then do the contrasts? I tried res_HA_PBS_vs_Infected <- as.data.frame(results(ddsMat, contrast=c("status", "PBS", "Infected"))) for the first comparison, but I'm not sure this is correct. Any suggestions would be appreciated!

Thank you!

rna-seq deseq2
• 2,062 views • link
updated by Ram Ram • written by AHerik • AHerik

Why use breed:status? Why not simply breed + status?

• 0 views • link
written by Ram Ram

@Ram, the design formula above would not enable differentiating between the effect of infection in different breeds for example but just give an "overall" effect of infection irrespective of different breeds.

• 0 views • link
written by Haci Haci

The Interactions section of the DESeq2 tutorial goes over this, I find this much simpler than using an interaction term and then trying to remember how to fetch results corresponding to different contrasts.

Initial note: Many users begin to add interaction terms to the design formula, when in fact a much simpler approach would give all the results tables that are desired. We will explain this approach first, because it is much simpler to perform. If the comparisons of interest are, for example, the effect of a condition for different sets of samples, a simpler approach than adding interaction terms explicitly to the design formula is to perform the following steps:

  • combine the factors of interest into a single factor with all combinations of the original factors
  • change the design to include just this factor, e.g. ~ group
• 0 views • link
written by Haci Haci

Does this mean just adding a new meta data column in which I have added the comparisons I want? For example, a combination of breed and condition so we have HA_infected, HA_uninfected, CA_infected, and CA_uninfected in the same place ?

• 0 views • link
written by AHerik • AHerik

Exactly. And you can convert this new variable to factor and set the levels if you like.

• 0 views • link
written by Haci Haci

I tried changing the factors to include only the comparison I wanted, but it led to NA values. And when I left all of the factors within the design:

resultsNames(ddsMat)
[1] "Intercept"                         "status_CA_PBS_vs_CA_Infected"     
[3] "status_HA_Infected_vs_CA_Infected" "status_HA_PBS_vs_CA_Infected"

which notably does not include all of the comparisons, such as "HA_infected_vs_HA_PBS".

• 0 views • link
updated by Ram Ram • written by AHerik • AHerik

Doesn't matter, you can have the code do any reasonable comparison you want it to, resultsNames is not a complete set of all possible comparisons among the factor levels.

• 0 views • link
written by Ram Ram

0 answers

No answers yet.

Log in to answer this question.

More posts like this

No similar posts found.