visualize CNV data based on WGS

Question

[Bogdan]

Dear all,

we have been calling CNV (copy-number variations) in cancer genomes based on whole-genome sequencing data (WGS). Please would you advise -- what would be the best way to visualize WGS in order inspect visually the CNV predictions ?
The BAM files of germline and tumor samples are big (> 60-80 GB).

thanks a lot !

-- bogdan

[Medhat]

Did you tried using terminal genome viewer like https://github.com/dariober/ASCIIGenome maybe it will help.

[Bogdan]

Thank you all for all your comments and suggestions ;)

Answer 1

[Pierre Lindenbaum]

I've written : http://lindenb.github.io/jvarkit/LowResBam2Raster.html , you're just limited by the memory...

it works for my needs, e.g:

.. found a familial deletion using lumpy-sv ... #ScienceItWorks pic.twitter.com/TSqGRxyEhF

— Pierre Lindenbaum (@yokofakun) January 12, 2018

Answer 2

[Leandro Lima]

Hi, Bogdan.

Have you tried IGV? You can load the BAMs and the BED files to check.

Leandro

[Bogdan]

yes, however we encountered a bit of problem with IGV : when zooming-out, the BAM tracks were not visible anymore in IGV.

Answer 3

[Eric T.]

You can generate a multi-page PDF of scatter plots focused on individual genomic regions (genes, chromosomes, etc.) with cnvkit.py scatter --range-list.

At UCSF you might be able to get access to the group license for Biodiscovery Nexus Copy Number, which is commonly used by pathologists for reviewing array CGH data. You can export CNVkit .cnr files to that program's format with the cnvkit.py export nexus-basic command, or include SNP allele frequencies from a VCF file with export nexus-ogt.

[Bogdan]

Dear Eric, thank you very much for your suggestions !

Answer 4

[clealk]

gw was designed for this purpose https://github.com/kcleal/gw

You can dynamically visualise whole chromosomes without memory problems

Answer 5

[NB]

You can try Alamut We use it and it works quite well for large BAM files.

Answer 6

[dariober]

Thanks to @Medhat for "promoting" my tool. However, also ASCIIGenome and any interactive tool working directly with BAM files will be slow when handling very large regions like CNVs (I have to check out Pierre's solution though!). In my opinion, best option is to convert BAM to bigWig or tdf and work with these instead. Have a look at deepTools (for BAM -> bigWig) and igvtools for BAM to tdf.

[Medhat]

It would be great If you could implement method to go directly to specific region in genome.

[dariober]

Hi- I think this is already available with the -r/--region option. To go directly to chr:start-end:

ASCIIGenome -r chr7:1000-2000 aln.bam

Is this what you mean? (By the way, ASCIIGenome, like IGV, will intentionally not show BAM tracks for very large regions).

[Medhat]

This is what I mean, My bad :)

[dariober]

No Worries- feel free to post questions, suggestions, bugs, whatever!