At which point should I do a log2 transformation of my iTRAQ LCMS data?

Question

[jolienhendrix]

Hi everyone!

I am currently analyzing proteomics data as part of my master's thesis but I am struggling with the log transformation. I need to analyze iTRAQ LCMS data but I do not have the original data that came out of Proteome Discoverer. I need to start from the iTRAQ ratios (so intensity of the protein in a sample/intensity of the internal standard). The final goal is to have the fold change expression ratios of the patients compared to controls.

My question to you is: at which point do I perform a log2 transformation? I have 2 options: - log2 transform the iTRAQ ratios --> calculate the mean abundance ratios per experimental group --> subtract the mean abundance ratios of patients with that of controls to obtain the fold change expression ratio - calculate the mean abundance ratios per experimental group --> divide the mean abundance ratio of the patients by that of controls --> log2 transform this ratio

Which option is best and why?

Answer 1

[Ahill]

You should take the first approach: log2 transform the ratios, calculate the mean, then subtract ratios. Reasons to prefer that approach are:

• log-ratios in data like iTRAQ should be more normally distributed than the raw-data, and so you can comfortably take a simple arithmetic mean of the log-ratios.
• having the log-transformed individual observations behind the means is often useful for visualization of wide-ranging ratios
• log-ratios are less subject to numerical round off issues that might happen with very large or very small ratios.

The final ratios from your second approach would be equivalent to the first if you did the geometric mean of the un-transformed ratios, but for the reasons above I'd use the first approach.