log/linear detection
Sara Ribeiro
I have a similar question. Previously, I used OmicsPlayground version 2 to analyze my proteomics data. Like Philipp, I receive data that has already been normalized according to the following method:
"Peptide relative peak areas were normalized to the total sum of fragment areas for the respective sample, and protein relative quantities were obtained by summing up to 15 peptides per protein."
Due to this normalization step, my values are extremely low. In version 2, this did not seem to be an issue, as protein abundance values remained positive. However, in version 3, I now receive an alert when uploading the abundance file stating:
"Log-transformed counts? OmicsPlayground expects linear intensities. Your data appears to be log-transformed. Would you like to undo the logarithm and convert to intensities?"
After accepting the transformation, the abundance (log2) values are negative, which I understand can happen when dealing with very low values.
I am aware that data processing was different in OmicsPlayground version 2, where values were transformed to log2CPM, a method more suited for RNA-seq than proteomics.
Howwever, how can I resolve this issue? Would obtaining intensity values before normalization be a possible solution?
Thank you in advance! And sorry, Philipp, for intruding on your question. However, since we seem to have doubts on the same topic, I thought it would be helpful to gather everything in one place.
Hi Sara. I guess the question is a bit different than Philipp's. In your case, indeed we tried to detect log/linear in v3 but sometimes it has problems determining that. Especially if all your (linear) values are between 0-100, it might think they are log transformed. If you are sure your values are linear (or sometimes people have percentage values) then do not convert "again" to linear. In that case you should keep the values as is. Hope this is clear. If you need further clarification, please open a new thread as this is more about logarithmic/linear conversion.
Ivo
Thank you for the explanation.
I followed your suggestion, but I didn’t notice any differences in the final values presented. I have one more question: Since my values range between 0 and 1, their log-transformed abundances are all negative. Could this pose a problem for subsequent analyses?