Question: How to exclude noisy CNV from study?
0
gravatar for rmateos.1
17 months ago by
rmateos.110
rmateos.110 wrote:

Hi,

Nowadays I am working with Whole exome sequencing data. After using Varscan2 and DNAcopy, I identified some samples as quite noisy:

Example of noisy sample

My plan is to exclude these samples, but I want to have an statistical proof for doing it. Would it be possible to receive advice about this topic?

Thank you in advance

ADD COMMENTlink modified 16 months ago by vchris_ngs4.5k • written 17 months ago by rmateos.110

Axes/color description could help intepreting the figure.

ADD REPLYlink written 17 months ago by Carlo Yague4.1k
1

You are absolutely right.

The green/black colors define different chromosomes. The red fragments are the segments obtained after applying the DNAcopy package to the output from Varscan2

The Y axis is the log ratio between normal and tumor.

The X axis is just an index putting all the segments in order. It is not the real length of each fragment in the chromosome, but more like the number of outputs obtained by the package

ADD REPLYlink written 17 months ago by rmateos.110

Thx for clarifying. I'm not very familiar with varscan and DNAcopy, but perhaps using longer segments could help stabilize the log ratio.

ADD REPLYlink written 17 months ago by Carlo Yague4.1k

Thank you for your advice. Other patients get very clear segmentation, and I know that the data is indeed noisy... I am still wondering if forcing longer segmentation in noisy samples is worth it...

Thank you again

ADD REPLYlink written 17 months ago by rmateos.110

Just a suggestion : you could try to use CNV-seq that does more or less the same kind of analysis, but calculate automatically the 'optimal' segment size based on the read coverage. That make sense because higher read coverage gives more power to detect differences over smaller segments.

ADD REPLYlink modified 17 months ago • written 17 months ago by Carlo Yague4.1k
0
gravatar for vchris_ngs
16 months ago by
vchris_ngs4.5k
Seattle,WA, USA
vchris_ngs4.5k wrote:

Just to add what Carlo said. There are other plenty of tools that you can use. DNA copy that performs the CBS in VarScan2 might not be the correct one for your data. If the depth is too high then the segmentation will create such noises. So you can use the longer segments. Or alternatively use methods like ADTex, Control-FREEC, Excavator. See how the results vary.

ADD COMMENTlink written 16 months ago by vchris_ngs4.5k
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 1741 users visited in the last hour