JISTIC. Make Probe location file to compare copy number among samples

Question

[Chirag Nepal]

Hi there,

I am trying to use JISTIC tool to analyze CNV data. I did not properly understand how to make the ProbeLocationFile.

Let's say, I have segmented file from DNACopy. The output files look like:

==> TestSample1 <==
barcode    chromosome    start    stop    mark    segmean
sampleS1    1    567331    569882    6    -2.3721
sampleS1    1    761847    905536    162    -0.1232
sampleS1    1    905636    3549990    2846    -0.0335

==> TestSample2 <==
barcode    chromosome    start    stop    mark    segmean
sampleS1    1    761857    7854094    4700    -0.0742
sampleS1    1    7858583    7861350    8    -0.8358
sampleS1    1    7862937    11905939    3063    -0.0598

==> TestSample3 <==
barcode    chromosome    start    stop    mark    segmean
sampleS1    1    569890    761867    2    -0.0945
sampleS1    1    761967    1231403    767    -0.6182
sampleS1    1    1231503    1233381    8    -0.1089

As, chr-start-end position in each file are different. How do we make the same region for all file. Because not all regions are present in all samples.

So, how to make "Probe location file"?

Could you please help me with this. Thank you for your time and consideration.

cheers

Chirag

[poisonAlien]

Just wants to make sure. Is it GISTIC you're asking about? If yes, I guess you're asking about marker file?

[Chirag Nepal]

I was referring to JISTIC http://www.c2b2.columbia.edu/danapeerlab/html/jistic.html , which is meant to be like an upgrade to GISTIC. I plan to try both. But for now, got stuck with making probe location file.

Answer 1

[Chris Miller]

You need to be using the data pre-DNAcopy segmentation. In the case of WGS, that'll usually be read depth in genomic windows and for exomes, that'll be depth in small sections of the covered regions (as output by something like Varscan exome CN calling). The locations of those windows are your probe location file, and your input data is the read depths or log2 ratios from each of the windows.

[Chirag Nepal]

Thanks Chris!

OK, will try that.

[Chirag Nepal]

Hi Chris,

I ended up using varscan (pre-DNA copy segmentation) and GISTIC (for population analysis).

ftp://ftp.broadinstitute.org/pub/GISTIC2.0/GISTICDocumentation_standalone.htm

Output from varscan

head P1.copynumber
chrom    chr_start    chr_stop    num_positions    normal_depth    tumor_depth    log2_ratio    gc_content
chr1    567331    567430    100    38.3    6.1    -2.515    47.0
chr1    567431    567459    29    29.8    5.6    -2.270    41.4
chr1    567497    567596    100    151.1    18.6    -2.882    48.0

For (6) Seg.CN (log2() -1 of copy number)

So, should I use: Seg.CN = (log2(-2.515 )-1) (why there is -1). In my trial run, I used log2(-2.515), and so on for each line.

I guess, as you suggested, it does not make sense to use output from DNAcopy segmentation, which are basically only the significant hits.

Thanks for your suggestions!