I ran an experiment with a ref gene (3 biological repeats x 3 technical repeats) and normalized gene expression of my target genes to the ref gene. I decided to present the data as log2 fold changes, which is, in my opinion, a better visualization of this kind of data (and of course, it helps with data normalization). Problem is, I can't find how to calculate the standard error of normalized, log transformed data anywhere. My solution was to calculate a 95% confidence interval and convert the range to log2. My question is, is this solution sound, and is it acceptable to present data in this manner? and how can I calculate the SE of this sort of converted data.
Thank you
The ddct value is the log-ratio of the (normalized) expression. The base of the log is the amplification efficiency. The efficiency should actually be (close to) 2.0, so this is the result you want to have. If you have estimated the efficiency to be different that 2.0 then you can change the base to 2 by deviding the results (estimate and limits of the CI) by log2(efficiency).
The SE for the normalized data is itself normalized accordingly according to this pdf:
http://www.pugetsound.edu/files/resources/qpcr-hand-calculations--relative-single-reference-.pdf
I can not use the same manipulation for log SE, as the Relative quantity of the ref gene when log transformed is zero.
The ddct value is the log-ratio of the (normalized) expression. The base of the log is the amplification efficiency. The efficiency should actually be (close to) 2.0, so this is the result you want to have. If you have estimated the efficiency to be different that 2.0 then you can change the base to 2 by deviding the results (estimate and limits of the CI) by log2(efficiency).
> Problem is, I can't find how to calculate the standard error of normalized, log transformed data anywhere
For normalized (log-transformed) data Jochen's steps 1-6 are right.
If You want to pesent SE(CI) for non-transformed data (return to source data) -have a look at http://www.amstat.org/publications/jse/v13n1/olsson.html
If one wants to use DDCt directly as log transformed data, you have to be careful as it can become upregulated to downregulated in graph.
The best is just get FC from DDCt and then convert it to log base 2 DDCt value. Here we use DDCt formula (DCt treated - DCt control) but not (DCt control - DCt treated).
Mainly, you want to compare your treated value with "0" (control) then you decide if you want your FC to log base 2 (FC). If not with FC only without converting to log 2 base, you must compare your treated FC value with "1" (control FC)
log base 2 (1) is "0"
- Badges
- Science topic