Entire TCGA mutation collection

[fh_fingolfin]

Hi all,

I'm a bit lost with TCGA. We are trying to make a global analysis of mutations in genes in cancers categorized per tissue type / cancer type. We thought about utilizing the TCGA data for this, so what I'd basically need is a global maf-file based on the entire TCGA data. However, it appears to me that there is no such file apparently available. I checked the TCGA data portal, but it seems it only lets you query cancer types one by one and then you still have to search for appropriate maf-files. Now I have come across Broad's Firehose portal, where they have a maf-file Dashboard which appears to be regularly updated. They list maf-files from several centers per cancer type, but also sometimes several files from the same center for a single cancer type. Does anyone know if the files listed there should in principal contain all TCGA mutations and I would get a global mutations file by simply combining (and removing duplicates) the maf-files from the Firehose MAF dashboard?

I take it, there is no other data source already containing all TCGA mutations like COSMIC?

Any feedback would be greatly appreciated - thanks!

[fh_fingolfin]

Bump - anyone?

[Richard Finney]

Anybody that can find and link to an official statement or an unofficial explanation on the status and quality of TCGA VCF and MAF file would be doing many a favor.


I suspect the methods of generating these files and the final quality vary greatly.

[Jessica_L]

I've been a bit lost when working with TCGA myself. As far as I can tell there isn't a global maf file, either.

It's my understanding that the Broad firehose data is built to both analyze and catalog analyses of TCGA data, so if I understand all of their About statements correctly, that means if you download the files from firehose and remove duplicates, you should have everything covered by TCGA.

[fh_fingolfin]

Thanks for the feedback. So it seems the whole TCGA project is quite a mess regarding data-presentation which is a pity. The COSMIC portal in contrast is neatly organized I find. I wonder why can't they do something similar - in the current state it's just a pain.

So it seems the conclusion so far is that downloading everything you can grab from Firehose, merging and removing duplicates seems to be the only feasible way to get a general TCGA mutations list. I'll go with that for now, but if anyone knows a better solution, let me and others know - I can't be the only one asking these questions.

[mjn138]

OncoLand/cBioPortal

There are a number of paid tools out there that have attempted to remedy the situation (by cleaning up the data, etc.). OncoLand, from Omicsoft, is one: http://www.omicsoft.com/oncoland-service, as well as free tools from cBioPortal (www.cbioportal.org).

[temizna]

As far as I know there is no global mac file for TCGA. cbioportal also only allows one cancer at a time download.

Why would you need a global file? You can download all the level3 maf files for all the available cancers and load them one by one or combine them by tissue later yourself.

When I work with TCGA mutation landscapes, I generally load one cancer type, analyze its mutation landscape save the results and move on to next one.
After this initial analysis is done I can then compare various cancers be it by type, tissue, sex whatever you can think of.

HTH,
Alpay

[fh_fingolfin]

Yeah seems cBioPortal only allows downloading a single cancer type and also only mutations for a subset of genes.

I'd need a global file just to avoid the hassle of searching for each cancer type, download its maf files and analyze/join the results. Having a global file is just way more convenient and apart from file size I don't see a particular advantage of doing it sequentially cancer by cancer.