Hi, I had a few questions about normalization for one-factor statistical analysis. I’m analyzing label-free Orbitrap proteomics data, with proteins in rows and samples in separate columns. Each sample column represents an independently acquired replicate/run.
Should I normalize the data before analysis, or is log2 transformation sufficient? I’m comparing inducer-treated samples with controls and expect higher protein abundance in the inducer group. I don’t think sum normalization is appropriate because it could artificially increase the relative abundance in the control samples. Would median normalization be more suitable?
Also, does MetaboAnalyst normalize each sample column separately, or does it apply normalization across the entire dataset?
