Thanks Henrik for the reply.
FYI.
I ran the analysis twice: once using a paired approach and once using
a large global reference group.
Method: CRMAv2 -> CbsModel -> GISTIC
There was approximately 50% agreement between the two results.
I noticed one interesting fact:
Almost ALL regions that were found when using the global reference,
but were *not* found when using the matched normal approach, were
regions known to contain common CNV -- (as reported by the Database of
Genomic Variance and nicely displayed on Nextbio's Genome Browser).
My tenuous conclusion is that when using the global reference the
naturally occurring CNV's in each individual overwhelm the top
results, hence hiding CNA that may be "associated" with the phenotype
of interest.
Cheers,
Greg
On Feb 12, 3:16 pm, Henrik Bengtsson <henrik.bengts...@aroma-
project.org> wrote:
> Hi.
>
>
>
>
>
> On Thu, Jan 27, 2011 at 11:03 PM,GregoryW<greg.d.w...@gmail.com> wrote:
> > Hello,
>
> > I was hoping to get some insight as to whether using a pooled
> > reference or paired normal is more appropriate when performing cbs.
>
> > I could see how using a paired tumor normal approach could be
> > beneficial, say, by negating benign CNVs in the individuals germline.
> > I believe these germline CNVs would have a better chance of being
> > reported significant when using a pooled reference.
>
> > However, I could also see how germline CNVs that may be indicative of
> > a predisposition to the tumor would be eliminated when using the
> > paired tumor normal paradigm.
>
> > Are there other factors I should be taking into account? Let's say
> > all my samples were performed from the same institution. Any
> > suggestions of how to compare which result is "better"?
>
> I would say it depends what you are looking for. If you are
> interested in the difference between a tumor and a match germline
> ("normal"), then you want to do a paired analysis. If you worry about
> the germline not being "normal" (i.e. diploid) everywhere, then you
> may also need to look at the normal relative to an "external global"
> reference (e.g. a very large pool of normals). Looking only at the
> tumor relative to an external global reference will never let you know
> whether a CN aberration observed in the tumor is somatic or in the
> germline. Thus, I'd say collection paired tumor normal data is always
> useful.
>
> Ideally, we'll one day have a CN segmentation and CN calling algorithm
> that operates on (tumor, normal, global reference) and provides
> easy-to-interpret results.
>
>
>
> > Thank you in advance,
>
> > Greg
>
> > P.S. Did TCGA use a pooled reference when finding aberrant regions?
>
> Sorry, I've been a little bit out of the TCGA loop recently, so take
> my comment on this with a grain of salt; A while ago when the ovarian
> (OV) analysis was mainly done on paired CN ratios. This was basically
> due to lack of resources/time, but the above issues were definitely
> discussed.
>
> Hope this helps
>
> /Henrik
>
>
>
>
>
> > --
> > When reporting problems on aroma.affymetrix, make sure 1) to run the latest
> > version of the package, 2) to report the output of sessionInfo() and
> > traceback(), and 3) to post a complete code example.
>
> > You received this message because you are subscribed to the Google Groups
> > "aroma.affymetrix" group with websitehttp://www.aroma-project.org/.
> > To post to this group, send email to aroma-affymetrix@googlegroups.com
> > To unsubscribe and other options, go tohttp://www.aroma-project.org/forum/
--
When reporting problems on aroma.affymetrix, make sure 1) to run the latest
version of the package, 2) to report the output of sessionInfo() and
traceback(), and 3) to post a complete code example.
You received this message because you are subscribed to the Google Groups
"aroma.affymetrix" group with website http://www.aroma-project.org/.
To post to this group, send email to aroma-affymetrix@googlegroups.com
To unsubscribe and other options, go to http://www.aroma-project.org/forum/
