Hi there,

Just a quick question regarding analysis of dose-control data. I've got paired cell data with 5 biological replicates: 5x controls, 5x dose1, 5x dose2, 5x dose3, 5x dose4, and 5x dose5. I've selected the most important features using multivariate tools on the entire data set with all doses. From here I can then run limma or another t-test to assign significance to the features.

However, I was wondering if it makes more sense to select the features (and subsequently perform univariate analysis) on the separate comparisons so e.g. control versus dose1 only.

I guess I am just wondering if when all clumped together into a single statistical run, I am missing out on features that may be uniquely significant to a specific dosage?

Any feedback on that is much appreciated.

Thank you!