This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Deseq2 with opposite (dominant negative) design

I am not sure this is possible but I was wondering if there was a way to model for opposite effects in the deseq2 design formula?

I have two cell lines, one overexpressing a Wt kinase and one a kinase-dead form. We have shown through experiments that the KD version has DN effects. Is there a way to model the differences between the two specifying their opposite effects?

I have done:

  • Wt kinase overexpressed vs not overexpressed control on it's own

  • KD kinase overexpressed vs not overexpressed control on it's own

  • Looked for opposite results between the two

Now is there a better way to model opposite differences between Wt vs KD with respect to their own controls in the DESeq2 formula?

would it be something like ~ cellLine + overexpression + cellLine:overexpression with 3 levels to overexpression?

rna-seq deseq2 r
• 910 views • link
updated by swbarnes2 swbarnes2 • written by jack.henry • jack.henry

Can you provide your full sample sheet?

• 0 views • link
written by rpolicastro rpolicastro

1 answer

Let's say you did the two comparisons you outlined above. You could take each gene, and subtract the WT fold change from the KD fold change, and the farther that is from 0, the more the gene response differed between the two gene versions. But you would have no p-value to go along with that difference in fold changes.

The design you outlined above will do that difference of fold changes, and give you a p-value. The column header of the results object will still say Log2FoldChange, but it will really be the difference of fold changes.

• 0 views • link
written by swbarnes2 swbarnes2

Log in to answer this question.

More posts like this