Question: It's there any GISTIC2.0 like tools?
0
gravatar for MatthewP
9 months ago by
MatthewP660
China
MatthewP660 wrote:

Hello, I re-edit this question, orginal question:
GISTIC2.0 can find CNV across a set of samples. I want to know it's there other tools can do the same thing? Thanks.

New question:
So GISTIC2.0 is more suitable for array CGH data, for example GISTIC2 requires markers file(-mk) which WES/WGS can't afford(We can build by-hand, but this making -mk useless, right?). It's there any tool implement GISTIC to WES/WGS data?

I use cnvkit to get cnv seg file by the way.

cnv • 277 views
ADD COMMENTlink modified 6 months ago • written 9 months ago by MatthewP660

but wait, GISTIC2.0 is not a CNV caller. It is a tool for driver genes discovery. Can you clarify what you need - cancer driver genes discovery or CNV calling? Why do you need another tool? Because you can not call variants with GISTIC2.0 or because you are not satisfied with the algorithm? I'd say GISTIC2.0 algorithm sounds bullet-proof for me.

ADD REPLYlink modified 9 months ago • written 9 months ago by German.M.Demidov1.6k

GISTIC calls CNVs for you, it just doesn't do segmentation. Though yes, it also does driver gene discovery, so a clarification would be nice.

ADD REPLYlink modified 9 months ago • written 9 months ago by jared.andrews076.2k

I am confused. Copy-number estimation is required by GISTIC as input (ftp://ftp.broadinstitute.org/pub/GISTIC2.0/GISTICDocumentation_standalone.htm , segmentation file paragraph). Clearly it requires Seg.CN as input. In which sense it calls CNVs?

ADD REPLYlink written 9 months ago by German.M.Demidov1.6k
1

Segmentation is breaking up the genome and estimating CN for each segment. However, these segments may be noisy and aren't stitched together into contiguous regions of CN change by segmentation algorithms alone.

GISTIC (and most other CNV programs/packages) utilize segmentation to perform CNV "calling" and to come up with discrete regions of CN change based on threshold values (often removing low-quality segments e.g. those with very few reads/probes). Often, a comparison to normals will also be done to remove germline changes, particularly if small focal changes are of interest.

This distinction is admittedly somewhat semantic, but generally the "calling" step is where hard thresholds are drawn and integer CN values are assigned. GISTIC does this.

ADD REPLYlink written 9 months ago by jared.andrews076.2k

Ah, I see. That's kinda weird - GISTIC has no idea about the underlying sub-clonal structure, so I'd never use it for calling, but good to know that it also does that, thanks!

ADD REPLYlink written 9 months ago by German.M.Demidov1.6k

you can try other algorithms: haar segmentation (CNVkit supports as I understand), CBS implemented in DNAcopy @ MatthewP

ADD REPLYlink modified 6 months ago • written 6 months ago by cpad011213k
2
gravatar for jared.andrews07
9 months ago by
jared.andrews076.2k
St. Louis, MO
jared.andrews076.2k wrote:

There are lots of them:

Are just a few off the top of my head. CNVkit is probably the easiest to use out of those and can work on arrays or WGS. There are many others you can find just by searching a bit.

ADD COMMENTlink written 9 months ago by jared.andrews076.2k

Sorry, I misunderstand function of gistic. I thought it's use to identify CNV from multiple samples, like some kind of merge many CNV results.

ADD REPLYlink written 9 months ago by MatthewP660

It kind of does that. It basically finds the recurrent CNVs between samples and tries to tie them back to which genes are affected.

ADD REPLYlink written 9 months ago by jared.andrews076.2k
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: 1039 users visited in the last hour