Hello,

I have a multifactor RNA-seq experiment I’m trying to analyze with DESeq2 (Vs. DESeq2_1.8.1). I am not sure how to setup the contrasts to find the right subset of deferentially expressed genes.

This is the design I use:

design = ~ treatment + genotype + genotype:treatment

Treatment factor has 2 levels: “Tret” or “Mock” (Mock is the base level)

Genotype factor has 4 levels: “A”, “B”, “C”, “D”.

My resultsNames(dds) are therefore:

[1] "Intercept"                                        " treatmentMock "            

[3] " treatmentTret "                               "genotypeA"                   

[5] "genotypeC"                                     "genotypeD"                   

[7] "genotypeB"                                     " treatmentMock.A"

[9] " treatmentTret.genotypeA"             " treatmentMock.C"

[11] " treatmentTret.genotypeC"            " treatmentMock.D"

[13] " treatmentTret.genotypeD"           " treatmentMock.B"

[15] " treatmentTret.genotypeB" 

I am interested in finding  genes with specific deferential expression in C+D genotypes and only in “Tret” condition.

Thanks a lot,

Moran

hi,

"differential expression in C+D genotypes and only in “Tret” condition"

Can you be very specific about what you are looking for? 

Do you mean, differences in genotype C or D vs A or B in Tret, and which are not differentially expressed C/D vs A/B in Mock?

With four genotypes, there are many possible ways there can be differences. There are six pairwise comparisons. Then there are situations where C and D are different than B but not A, etc.

Hi Michael, thank you for the reply

The structure of the experiment and aims in this analysis are to find those gene that are common to C and D and have differential expression compared to Tret and A and B. You can imagine that C and D share a genetic component (and so do A and B) and we want to find its influence on Tret

Thank you for reply and all the good work. I do see this is more of a general statistical question rather then a DESeq2 issue.

MF