Hello everyone,

I apologize in advance if this question may seem trivial, but I recently started working with proteomics data and I have some doubts. Is data normalization in log2CPM exclusive to RNA-seq data, or can it also be applied to proteomics data?

Thank you.

The term 'logCPM' is short for log counts/million counts, which is generally only applicable for count-based applications like RNA-Seq. Usually proteomics data are the area under the curve of the m/z for a given peptide as it elutes from the LC, and then summarized across all peptides for a given protein. In that situation you would likely want to take logs (I always do), but that wouldn't be considered logCPM because there are no counts.

I have seen instances where the proteomics data are instead counts of m/z 'hits', but that is pretty rare. I suppose you could use logCPM in that case, but do note that the rationale for computing logCPM in the context of RNA-Seq is to remove the inherent right skew as well as perform a shift-normalization, on the assumption that the total number of counts for a sample is an accurate reflection of the (variable) amount of cDNA applied for a given sample. It may also be true that you have counts of m/z hits, and additionally it may be true that the total count of those hits is proportional to the total amount of protein injected onto the LC, but that's an assumption you would have to make/defend.

In addition, your question is off-topic. General questions like this should be directed to biostars.org