I have 2 questions regarding elimination of samples:

1. We included 9 endogenous control genes in our qPCR setup (TaqMan
low density arrays, 32 genes run i triplicate) used for expression
profiling during a differentiation study (15 time points). In one of
the samples only one replicate was obtained for a particular control
gene. Subsequently we eliminated that control gene before the final
evaluation (in geNorm) of which genes to use for normalization. Since
all further enterpretation of data rely on the choise of control genes
we guess that this would be the most correct way to do it? The mean
M-value in geNorm for the 3 genes we ended up with was 0.2355 and the
paiwise variation below 0.10 - which is excellent.

2. After normalization of the data, again at one time point we had a
target gene with only one replicate (of three) coming up - however
since the expression level of the target gene can be seen in relation
to time points before and after and fits well in curve over time, we
guess it is ok to include (mentioning of course that the value
respresents only one replicate) - or? Otherwise we normally say that 2
of 3 replicates should be present with a difference in CT below 0.5
(CT´s below 32 - for CT´s above a difference of 1.0).

We have had MANY discussions in our lab how to deal with these things
and would deeply appreciate input from other users.