I need to calculate fold change (and log fold change but I know how to do that) manually (in excel) using normalized alignment counts (counts per million). I don't want to use DESeq2 or another differentially expressed gene application. I want to understand the math. I may be a bit masochistic, but can anyone walk me through it?
Thanks much,
cpm is a measure of concentration. You haven't specified the measurement technique used to generate the concentration: MS (TOF, Trap, etc) or scintillation counter. Like any concentration term the numerator is the represents to number of counts associated with the specific article being measured. This is an amount present in the measurement as opposed to a concentration. The denominator (per million) is a million counts of all the things in the sample being measured. The counts may be associated with a specific mz peak in a mass spectrum and the denominator the sum of the counts associated with all peaks in the mass spectrum. The problem is you need to know if the spectrum has been baselined or not for the total counts in the spectrum and the alternative spectrum that you are comparing it to covers the same mz range and has been baselined the same way. If that is true, then the ppm value represents the fraction of the total spectral counts that are included in the peak of interest. The conversion to ppm allows you to cross compare spectra of samples at different concentrations or different spectral accumulation times since the spectrum has been normalized to the total counts present. Violate any of the above assumptions, however, and you can't do a valid fold-change comparison.
Once you are certain what the numerator and denominator represent in the ppm concentration, and have ascertained that they represent the same things in each value to be compared, then you are ready to calculate the fold change.
Fold change = ppm of sample 1 / ppm of sample 2
Log fold change = Log (Fold change) = Log (ppm 1) - Log (ppm 2)
Log fold change normally means Log base 10 (Log10). This provides an order-of-magnitude estimate of the change (i.e., Log change = 2 is a 10^2 or 100-fold change in the relative abundance of the component between the two samples.
- Badges
- Science topic