Skip to end of metadata
Go to start of metadata

DCC Archive Submissions

Each pipeline executed by the BROAD TCGA GDAC Firehose pipeline results in a set of 6 files being submitted to the DCC: primary results in the Level_* archive; auxiliary data (e.g. debugging information) in the aux archive, tracking information in the mage-tab archive; and an MD5 checksum file for each. In most cases you will only need the primary results in the Level_* archives. Microsoft Windows-based users can use the WinRAR utility to unpack the archive files, while Unix and Apple Mac OS/X users can use the gzip and/or tar utilities. As of January 2013 our archives follow the nomenclature given below.  Look here for older version.

<Domain>_<DiseaseCohortName>.<TaskName>.<DataLevel>.<Runcode>.<Revision>.0

 

Element

Description of Permissible Values

Domain

the literal string gdac.broadinstitute.org

DiseaseCohortName

a string of the form

<disease_specification>[<sample_type>]

The disease specification most often refers to a single disease study given by its TCGA disease abbreviation, such as GBM for Glioblastoma Multiforme;  but may also refer to an aggregate of multiple diseases, such as PANCAN12 (which refers to a cohort of 12 diseases created to study pan-cancer trends) or COADREAD (which combines the single diseases COAD and READ into one cohort).

The optional <sample_type> suffix consists of a literal dash followed by a TCGA short letter code designating the tissue sample type; for example, the suffix "-TP" indicates that the given archive contains results based upon primary tumor data.  As a final example, here's how TCGA short letter codes  would most commonly map to sample sets in Firehose, for a single disease study:

Sample Set NameDescription
BLCAall tumor and normal samples for Bladder Urothelial Carcinoma (union of everything below)
BLCA-TPonly primary tumor samples
BLCA-TMonly metastatic tumor samples (if any)
BLCA-TRonly tumor recurrence samples (if any)
BLCA-NT

only tissue normal samples (if any)

BLCA-NBonly blood normal samples (if any)

TaskName

Tasks should be named as

<Datatype>_<AlgorithmName>

For example: CopyNumber_Gistic2. The datatypes correspond to columns 2-12 in any of our sample data tables

with several types spelled out in longer form for clarity as follows:

Short FormLong FormDescription
CNCopyNumberSNP6 copy number data
LowPCopyNumberLowPassLow pass DNASeqC copy number data
MAFMutationmutation calls

DataLevel

the literal strings Level_2 or Level_3 for stddata tasks, or Level_4 for analyses tasks

Runcode

10 alphanumeric characters representing the date and a unique "pass" identifier, such as 2011072800 to indicate "pass 0" over the July 28,2011 data snapshot;  or 2011072801 to indicates "pass 1" over same dated snapshot

Revision

a small integer (usually single digit) indicating how many times the given TaskName was successfully run in the given pass

 

 

Labels: