The problem here is that there's an absence of data for one of the genes
you want to compare, so one can't really assign any significance to that
particular gene or transcript.
This would require inventing a singificance value where if a gene is
highly expressed in one sample, but not measured at all in another, then
that gene should be said to be significantly differentially expressed.
If you are looking for situations like this I see no reason why you can't
create a rule for identifying them and pulling them out of the cuffdiff
Even then, you'll have to make some decision about what FPKM value would
be significantly different from 0 in another.
Hope this helps.
Dr. Graham Etherington
Bioinformatics Support Officer,
The Sainsbury Laboratory,
Norwich Research Park,
Norwich NR4 7UH.
Tel: +44 (0)1603 450601
On 07/09/2012 16:16, "suzan katie" <suz.katie(a)gmail.com> wrote:
I am comparing two samples (control and treated) paired end RNA Seq data.
In the cuffdiff output I have noticed that few genes have zero FPKM value
in one sample and other sample has significant FPKM value.
I want to identify uniquely expressed genes identified only in one sample
(either control or treated).
My question: If something is measured with significance in one sample
(high FPKM), but not measured at all in another sample (Zero FPKM),
should I consider that gene as significant?
Can anyone explain this.