[BioC] presence/absence call using oligo array?

[Sean Davis]

On 1/13/06 3:07 AM, "xpeng" <xpeng at utk.edu> wrote:

> Dear Bioc,
> 
> Has someone ever used long oligo arrays to test if genes were expressed or
> not? I have two microarray slides on which oligonucleotides (65mer) for genes
> from a parasite genome were spotted. One gene has one oligo but spotted twice
> on each array. One slide has a RNA sample extracted from infected tissues
> (therefore mixture of both parasite RNAs and RNAs from host tissues)
> co-hybridized with RNAs from uninfected host tissue (no parasite RNAs).
> Another is a biological replicate but with dye-swap. People want to know what
> parasite genes are expressed in the infected tissue. Is it possible to do so?
> Related publications are highly appreciated.

You cannot answer that DIRECTLY with this design, I don't think.  You could
look for differential expression between the two samples--that is probably
as close as you can get.  You can perhaps say what genes are NOT expressed,
as those spots will have very low intensity (within the range of
background), but that doesn't imply that the others are expressed.  And, you
might think that all genes that show some signal AND are differentially
expressed between the two samples would be expressed parasite genes, but
these could also be due to regulation of the genes in the infected host
combined with some cross-hybridization.  So, I would look for
differentially-expressed genes and see what that can tell you.  With only
two samples, doing meaningful statistics will be difficult, but you could
try using limma or some other moderated testing procedure designed for small
samples (and able to account for your replicated probes).  Alternatively,
just rank the genes by fold change.  In the end, if you REALLY want to know
which genes are differentially expressed, another couple of arrays would
probably help.

Sean