Hi dear all, I am trying to compare my RNA-seq data sets.how can I compare expression profile between groups as following: Case treated, Case untreated, Control treated, Control untreated. The p value <0.05 and fold change>2 of each comparison are not as the same and I miss my target genes when I filter each group.For example p val of TGF b1 in case treated vs. Case untreated group is 0.04 but in control treated vs.control untreated group id 0.1 and almost p value of the genes obtained in case treated vs.case untreated are different from control treated vs. Control untreated. Is it possible to select only one comparison for example case treated vs. control untreated and all following analysis perform on that?what is the main way to select an influenced gene after treatment? I want to select 3 target genes involved in a particular signaling pathway for qRT-PCR validation, but p value of each comparison is not as the same and I am confused . I would be so appreciated if you help me. Thanks
Hi! It seems that you have Case vs. Control and Treated vs. Untreated as your groups. What question are you trying to answer?
1) Are you trying to see what genes are differentially expressed in treated compared to untreated? You'd have a fold change: [treated] / [untreated].
2) Are you trying to see what genes are differentially expressed in case vs. control? You'd then have a fold change: [case] / [control].
3) Are you trying to see determine the fold change [treated] / [untreated], AND see whether that fold change is significantly different between case and control? In this case, you're testing a fold change of a fold change, like ([case_treated]/[case_untreated]) / ([control_treated] / [control_untreated]).
If you're doing number 1 and/or 2, it's basic RNA-seq analysis where you get a fold change and an adjusted p-value, and you select your "interesting genes" based off a cutoff for the fold change and the adjusted p-value.
If you're doing number 3, this involves a bit of a more complex design because you have two variables ('treated vs. untreated' and 'case vs. control') instead of just one. Here's an example of how one is done (similar to your example, here, there are two variables: A vs. B and treated vs. untreated): https://support.bioconductor.org/p/69705/ (From here, you should be able to get the fold change and adjusted p-values).
Hope this helps!