Thanks for code, I like it. First I confess to not fully reading the paper, I see it does a staunch defense of itself, but need more time to look at the details.
I had to read the simulation vignette a few times, I couldn't get what you were up to.
I gather now that you decide to treat the usual biological explainer variables (variable of interest) as batch effects and so subject to collection (well, modelling) in the surrogate variables sva() function. And then impose the thin_2group() procedure. It's a great idea, but I think mentioning the explainer variables as Batch Effect or Unwanted Variance might help set the scene better.
Also at the end, you say SVA2 "gets at" the original biological explainer variable, which is interesting but I would have thought it actually "catching it" because that is what we want in this simulation context. Of course it's very interesting because it's clear SVA can sometimes lead to stealing signal from the variable of interest, usually quite undesireable. I look forward to finishing reading your paper!
Thanks for code, I like it. First I confess to not fully reading the paper, I see it does a staunch defense of itself, but need more time to look at the details.
I had to read the simulation vignette a few times, I couldn't get what you were up to.
I gather now that you decide to treat the usual biological explainer variables (variable of interest) as batch effects and so subject to collection (well, modelling) in the surrogate variables sva() function. And then impose the thin_2group() procedure. It's a great idea, but I think mentioning the explainer variables as Batch Effect or Unwanted Variance might help set the scene better.
Also at the end, you say SVA2 "gets at" the original biological explainer variable, which is interesting but I would have thought it actually "catching it" because that is what we want in this simulation context. Of course it's very interesting because it's clear SVA can sometimes lead to stealing signal from the variable of interest, usually quite undesireable. I look forward to finishing reading your paper!