Hi,
On Mon, Dec 12, 2011 at 1:15 PM, Greg Wall greg.d.w...@gmail.com wrote:
CORRECTION:
ref.theta - extractMatrix(Normal)
theta - extractMatrix(Tumor)
C - 2 * theta/ref.theta
On Mon, Dec 12, 2011 at 1:09 PM, Gregory W greg.d.w...@gmail.com wrote:
Hello,
Many thanks for the site and quick feedback to the discussion board.
I was hoping to get a little clarification about the difference
between these two statistics: Raw CN and CBS mean.
After running CBS I get the following data frame:
cbs - CbsModel(tumor, normal, min.width=5, alpha = .01)
fit(cbs, min.width=5, verbose=0, force=TRUE)
regions - getRegions(cbs)
regions - regions[[1]][,1:5]
head(regions)
chromosome start stop mean count
1 1 51599 25455929 0.0870 14665
2 1 25465716 25519534 -1.2750 29 ***
3 1 25519574 26308518 0.0994 369
4 1 26313794 27104892 0.4272 336
5 1 27108799 38042779 0.1022 6161
6 1 38044082 38063269 -0.6952 8
Where column mean is the result from DNAcopy segmentation for the
particular region -- which is just the mean log-ratio-value of all
probes in the region. And I'm assuming that since I ran doCRMAv2 on
the data, the CBS code above is performed on normalized tumor and
normal data.
Focusing on regions 2 show above with trailing ***, if I run:
ref.theta - extractMatrix(ref.average)
theta - extractMatrix(file.group)
C - 2 * theta/ref.theta
And then calculate the mean raw CN using matrix C for the same probes
in regions 2 used to calculate the CBS mean of -1.2750: does this CN
mean relate in any way to the CBS mean? I realize the log ratio is
logged and centered around 0 for the segmented data, whereas, CN is
around 2, but aside from simple transformations can you help me
understand the difference between these two statistics?
There are two CN metrics here:
non-logged CN ratios: C - 2 * theta/ref.theta
log2 CN ratios: M - log2(theta/ref.theta)
Ignoring non-defined signals, the bijective relationship is M -
log2(C/2) and C - 2 * 2^M.
The CbsModel segments M (log2 CN ratios) and reports the mean values
on these M values per segment. The reason for this is because that is
what is used in the CBS papers.
In the Aroma framework, we try to represent everything on the original
scale, because that way we avoid issues of that log of non-positive
values etc. Thus, 'theta' and 'ref.theta' are on the non-log scale.
Moreover, and maybe less of value to this explanation, but personally
I prefer to think of CN data on the non-log scale, especially since we
can indeed have zero or very small copy numbers and then M is
undefined or minus infinity.
Does the mean CN for probes included in a particular CBS segmented
regions have any value?
I don't understand this question.
/Henrik
Many thanks!
Greg
--
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/
--
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/
