Hi,

I have some self-self design-based microarray data, and am unsure how to
process them as I am more accustomed to reference design-based data.

Briefly, I have 4 columns of log2-transformed expression values i.e.,

First array: cy5 of Treated sample, cy3 of Treated sample
Second array: cy5 of Control sample, cy3 of Control sample

Couple of questions:
A) Technically, is there 1 or 2 replicates ?
B) If there is 1 replicate, the intensity ratio of gene A is given by
0.5*[cy5(T)
+ cy3(T)] - 0.5*[cy5(C) + cy3(C)] right ?
C) Would it be meaningful to perform t-test by assuming 2 replicates such
that the intensity ratio of gene A (first replicate) is given by cy5(T) -
cy3(C) and the second is given by  cy3(T) - cy5(C) ? or the first is given
by cy5(T) -  cy5(C) and the second by  cy3(T) - cy3(C).
D) Some rows contain at least a missing value, should I remove those rows
before doing the averaging/substration and clustering ?

Other question:
A) I also received some .gpr, are there any how-tos to check what sort of
normalization is needed, and how to normalized the data ?

Thanks

	[[alternative HTML version deleted]]