R help - Dec 2010 - Statistics of count data for differential expression

Hi All,

This is more of a general statistics question that I am not so sure about so
I need some thoughts from you experts out there. I have Chip-Seq read count
data for 2 different treatments (T1 and T2) and my final aim is to calculate
differential expression between them. I also have Chip-Seq read count data
from a "control / wild type (C)" experiment. I have 3 replicates (2
technical, 1 biological) of each of these. What is the best way to go for a
differential expression between T1 and T2? Is it best to just go straight
between T1 and T2 or is it to do a T1vsC and T2vsC and then compare the
results from these for diff expr? What do you think is better and why? I
appreciate your thoughts on this.

Sincerely,

Vishal

	[[alternative HTML version deleted]]

Vishal Thapar <vishalthapar <at> gmail.com> writes:
> This is more of a general statistics question that I am not so sure about
so
> I need some thoughts from you experts out there. I have Chip-Seq read count
> data for 2 different treatments (T1 and T2) and my final aim is to
calculate
> differential expression between them. I also have Chip-Seq read count data
> from a "control / wild type (C)" experiment. I have 3 replicates
(2
> technical, 1 biological) of each of these. What is the best way to go for a
> differential expression between T1 and T2? Is it best to just go straight
> between T1 and T2 or is it to do a T1vsC and T2vsC and then compare the
> results from these for diff expr? What do you think is better and why? I
> appreciate your thoughts on this.
  Offhand I don't see why you need to involve the wild type if you
are really interested in differential expression between T1 and T2, but
I think you might get a more informed answer on the Bioconductor list ...

  Ben Bolker