TCGA MAF files after migration to GDC

Question

[nilus1432]

Where to find the latest recommended MAFs for TCGA? particularly after TCGA data has been moved to GDC.

1. There are no MAF files on the GDC,
2. The NCI wiki lists set of current and obsolete maf files, but the links mentioned are not working now. (Also in archive many files which are listed are not available for e.g. SKCM only one somatic file is present in archive)

I read some of the threads in BioStars which recommend some sources I have questions on these..

1. The Ding lab at TGI, WashU track the best available MAFs in this spreadsheet for use in MuSiC analyses.

Q1: The links for these files are no longer working

1. For Broad's firehose recommended files,

Q1: Why there are multiple files for certain cancers such as COAD, READ etc??

Q2: The links for these files are no longer working

1. Recently, CGC also has a set of recommended files,

Q: The links for these files are no longer working

Among the above three sources which files should be considered for analysis of somatic mutations? For, some cancer types, files are same among three above recommendations. But for cancers like ACC, firehose recommends different file than Ding lab and CGC?

Any directions for using right set of files from TCGA??

[Zhenyu Zhang]

"There are no MAF files on the GDC" This is wrong. All TCGA MAFs are in GDC, just in the legacy portal.

GDC has also generated protected MAFs from 4 different pipelines. In the June release, you can find them in the release notes; in the most recent release, they are searchable in the portal. For public MAF, only MuTect2 calls are currently available, the rest need additional germline filtering before they go public

Answer 1

[Andrew J. Rech]

The most complete high-quality MAF files currently available are the Broad 2016 spring run, which is available for each disease cohort: Firehose

In my opinion, working with the most recent Broad pipeline MAFs is the best use case for standardized MAF input to whatever analysis you are hoping to perform, but YMMV and keep in mind the discontiguous nature of TCGA disease cohort MAF processing / quality control when interpreting results.