qPCR analysis with error bars

Peter Ellis via methods%40net.bio.net (by pjie2 from cam.ac.uk)
Thu Dec 1 20:04:44 EST 2011


>
>   6.86±0.17 sure doesn't look like an arbitrary constant.   It looks like a measured
> value with uncertainty that will affect the outcome of the calculation.

Yes.  The ±0.17 is the uncertainty in your reference (untreated) sample 
though, not in the test sample.  You should be plotting both in your 
final figure, with the appropriate uncertainty on each.

So, in page 58 Step 5, after calculating ddCt, i.e. calibrating to the 
untreated sample, for you have:

Drug A:    -2.5 ±0.10
untreated:  0.0 ±0.17  <-- NOTE THAT THIS ZERO STILL HAS AN UNCERTAINTY

This is all clearly shown in Table 11, along with the figures for the 
other two drug treatments.  One place confusion may be coming in is that 
for some reason Table 11 and the subsequent workings on pages 57-59 use 
different figures for the s.d. for untreated/GAPDH.  Specifically, in 
table 11 this is given as ±0.09, whereas on page 57 (final line) they 
use ±0.022.  Carrying this difference forward is why page 59 line 3 
gives the untreated dCt value as 6.86±0.17, compared to the value of 
6.86±0.19 shown in table 11.

In all other respects the workings are accurate.  The uncertainty in the 
untreated sample stays with the untreated sample final value, the 
uncertainty in the Drug A sample stays with the Drug A final value, and 
so on.

Peter


More information about the Methods mailing list