Trying to understand edgeR logFC and logCPM

[sindrle]

Hi!
Im trying to calculate standard deviation for the Fold Change, or logFC, in edgeR manually.

This is the result for one gene using edgeR:

logFC logCPM
-0,393306318 0,839097928

Im trying to reproduce this result manually.
Im trying to calculate the logFC manually and then get a standard deviation.

First, I got all the individual logCPM and CPM from my samples.

I then calculated: (logCPM group 2) / (logCPM group 1)
I also tried: (CPM group 2) / (CPM group 1)

The result was 1,546904843 and 1,348790178 ... Somethings not right..
I then took log(1,546904843;2) = 0,629384453... Still far off...

I did the same for CPM, took log(1,348790178;2) = 0,431665935...

So, how is these values calculated? And is it possible to get a measurement of individual variability?

[sindrle]

I might add that the same calculations on Cuffdiff 2.1.1 output works like charm! Even though its useless...

Only problem is that YOU DONT KNOW WHICH VAULE IS FROM WHICH BAM!!!!

http://seqanswers.com/forums/showthread.php?t=39978