Apologies in advance if this is a basic question, I think I’m making it out to be more complicated in my head than it needs to be. I wanted to make sure I was approaching it correctly:
I have two experiments with a treatment vs wild-type. Both were performed similarly, but one experiment was performed with a knockout gene and the other was not. So far I have performed a t-test within each experiment to test the outcome of the protein in treatment vs wild-type. My question is, would it be correct to compare the ratios of each experiment to the other through an odds ratio and then covert that to log odds and create a volcano plot? I want to look to see if their are differences in the protein expression outcome between the two experiments.
Hopefully this makes sense and I’m just over-complicating something that doesn’t need to be.
Thanks in advance!
Was WT one experiment and treatment one experiment, or do you have two WT vs treatment? Is this mass spec, if so I would strongly suggest to use a dedicated statistical framework such as DEqMS or DEP from Bioconductor rather than a simple t-test.
I would have two WT vs treatment (one for each experiment) except that one experiment would also have a Knockout Gene and the other would not) - thereby would be a 2x2 comparison I believe.
DEqMS seems like a nice plan, however dealing with duplicate proteins in my rownames might be an issue that still does not have a proper solution aside from just making them unique IDs... hence why I am normalizing via log2 and then running a t.test..