cda-meeting-notes-07-14.md

• Is there a schema describing how data in each table links to each other?

  • case barcode - covers 90% of linkages
  • sample barcode / aliquot barcode cover rest
  • aliquot - one sample split into multiple test tubes
  • DNA methylation tables: Different chromosomes / different human genomes

• Significance of version numbers on tables

  • Versioned tables are kept around so that code that references one version of the data can be run in the future w/o breaking

• What’s the use case for the REST API

  • It's meant for the website

• Within just TCGA + CPTAC, are there any data inconsistencies that the CDA team has already discovered that we need to watch out for

  • Contact them as they pop up

• submitter_id

  • case_barcode and submitter_id are generally the same (from GDC)
  • submitter_id can sometimes be ambiguous
  • GDC barcodes: begin with case barcode, +4 letters for sample, +4 for aliquot

• Methylation tables are split to lower query costs for end users

• Tables are kept flat - can be represented in a spreadsheet

• mRNA seq - one file per case per experimental run

• Billing account for external users, contact Fabian

• The metadata associated with a sample is always kept current in the _current tables

  • We should use the tables that are marked _current
  • Data in the current tables change when GDC runs a new harmonization pipeline