Question: How To Process The Cnv Level 3 Data From Tcga
gravatar for J.F.Jiang
7.0 years ago by
J.F.Jiang820 wrote:

Hi all,

as we know, TCGA offered us a good resource for genomic analysis. Here I have a question, we should subscribe the normal segment score to obtain the somatic value.

However, how to process this approach is quite confusing.

Because the segment region for tumor and normal different quite a lot. So is there any hints for this kind of data?

And another question is that, for CNV data obtained from SNP affy arrays, TCGA offered cnv.seg and nocnv.seg for one sample, so which one should I use? suppose should be cnv.seg, so what's the usage of nocnv.seg file?


tcga cnv • 15k views
ADD COMMENTlink modified 7.0 years ago by Ryan D3.3k • written 7.0 years ago by J.F.Jiang820
gravatar for Ryan D
7.0 years ago by
Ryan D3.3k
Ryan D3.3k wrote:

The TCGA data is an incredible resource, but knowing how to navigate it can be tricky. The level 3 data provides a mean copy number estimate of segments covering the whole genome (inferred from Affy SNP 6.0) A paper studying eQTLs in breast cancer adjusted for expression of each gene as determined by this data. The paper is called "Integrative eQTL-Based Analyses of Breast Cancer" and was published in Cell. You may see the method which they used.

For expediency, however, I can point you to a file which has estimated copy number of the gene that was used in the paper. You may use it to see if the level 3 breast cancer CNV numbers and the authors' own estimations agree. The Level 3 TCGA data should just have the segment mean of the coding region of the genes in this file. Given that the measure used is log ratio-based, I think 0 should correspond to 2N.

ADD COMMENTlink modified 7.0 years ago • written 7.0 years ago by Ryan D3.3k

thanks, i have read this paper before, however, their process to CNV is just a coarse method to transform the level 3 data. I think for more specific, i should look into RAE to build the CNV across each gene, however, i still can not obtain the same data as TCGA offered. Asked the author, however, no response back.

Another question is that when we deal with level2 or level 3 data, which have been already normalized, should we use the tumor intensity - matched normal intensity to obtain a tumor-over-normal one? I suppose TCGA has already done this process?

If you know, could you just give me some points?

And of cource, thank you for sharing the data in the paper you mentioned, though may be rough, but could be quite useful to me. Send the file to if you can,


ADD REPLYlink written 7.0 years ago by J.F.Jiang820

FYI, File is something like 200Mb+ unzipped. I will email you info so we can set up transfer.

ADD REPLYlink written 7.0 years ago by Ryan D3.3k

thanks, great help

ADD REPLYlink written 7.0 years ago by J.F.Jiang820

I also read this paper. I am trying to understand how to retrieve the segmented copy-number scores of the tumor samples and the paired-normal control from the level 3 data. Anyway can help me out? Thanks.

ADD REPLYlink written 6.5 years ago by liu4gre200
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 1836 users visited in the last hour