possible CNVineta bug / strange behavior
16 views
Skip to first unread message
Matt Shirley
May 20, 2011, 4:16:19 PM5/20/11
to CNVineta
I am using CNVineta to analyze copy number results from PennCNV, and
have loaded BAF,LRR,and annotations along with the segmentation data
to create a binary file. The binary file loads excellently, and both
the dox0 and log regression results seem logical. My question is this:
When I specify a locus:
chrom = "3"
start = 163995351
end = 164108689
and do:
visualizeRegion(CNVdata = objCNVineta,
batch.case=c(case),batch.control=c(control), chrom = chrom,
start = start, end = end)
I end up with a plot with a number of CNVs found in case or control,
along with some statistical tests. This is great, but the plot and
statistics do not seem to be only for the region that I specify. For
the above example, the plot/statistics are for the region from
163910721 to 164177652. Below is a link to the plot that shows the
issue. In my mind, the plot should only be showing me 2 CNVs in cases
and 0 CNVs in controls.
An example of the output mentioned above: http://goo.gl/KyCPP
have loaded BAF,LRR,and annotations along with the segmentation data
to create a binary file. The binary file loads excellently, and both
the dox0 and log regression results seem logical. My question is this:
When I specify a locus:
chrom = "3"
start = 163995351
end = 164108689
and do:
visualizeRegion(CNVdata = objCNVineta,
batch.case=c(case),batch.control=c(control), chrom = chrom,
start = start, end = end)
I end up with a plot with a number of CNVs found in case or control,
along with some statistical tests. This is great, but the plot and
statistics do not seem to be only for the region that I specify. For
the above example, the plot/statistics are for the region from
163910721 to 164177652. Below is a link to the plot that shows the
issue. In my mind, the plot should only be showing me 2 CNVs in cases
and 0 CNVs in controls.
An example of the output mentioned above: http://goo.gl/KyCPP
Michael
May 23, 2011, 4:40:28 AM5/23/11
to CNVineta
Dear Mr. Shirley,
Thank you for reporting that issue. You are right, the tool was
counting all CNVs within the plot and not only those overlapping the
area which the user requested the plot for. I changed this for version
1.0-1 which is already available. The plot caption still shows the
number of CNV carriers for the whole picture/plot. But the segment
stats are now calculated for the region of interest.
If you are wondering why the plot is often larger than the requested
area, we extend the plot to the outer limits of the largest CNVs
overlapping the region of interest. Sometimes you will find regions
which are overlapped by a very big CNV, which causes CNVineta to zoom
out extremely. If you find a plot like this, you can overcome this
problem by setting proper value for "max.extension=".
Kind regards,
Michael Wittig
Thank you for reporting that issue. You are right, the tool was
counting all CNVs within the plot and not only those overlapping the
area which the user requested the plot for. I changed this for version
1.0-1 which is already available. The plot caption still shows the
number of CNV carriers for the whole picture/plot. But the segment
stats are now calculated for the region of interest.
If you are wondering why the plot is often larger than the requested
area, we extend the plot to the outer limits of the largest CNVs
overlapping the region of interest. Sometimes you will find regions
which are overlapped by a very big CNV, which causes CNVineta to zoom
out extremely. If you find a plot like this, you can overcome this
problem by setting proper value for "max.extension=".
Kind regards,
Michael Wittig
Matt Shirley
May 23, 2011, 11:11:08 AM5/23/11
to cnvi...@googlegroups.com
Great! I would like to download this new version, but can't find it in the website. Could you point me to it?--
Matt Shirley
Ph.D Candidate - BCMB
Matt Shirley
Ph.D Candidate - BCMB
Matt Shirley
May 23, 2011, 1:09:50 PM5/23/11
to cnvi...@googlegroups.com
Never mind. I did a little URL replacement surgery and got it. Thanks,
Matt
Reply all
Reply to author
Forward
0 new messages