JSON revisions before preprocessing freeze

[Kcjohnson]

As we have begun to preprocess and align, a need for revising some of the json fields has arisen. We intend to re-run all samples through the pre-processing and alignment pipelines while making the following changes to the json files:

• [x] Edit ID tag to include ~full flow cell name~ first five characters of flow cell ID
• [x] Include current missing information in file size and file md5sum
• [x] Replace TCGA/HF/JDG readgroups with the same format as GLASS-WG
• [x] Include aliquot UUID in RGSM tag
• [x] Revise pair ID to incorporate both tumor and normal ID (rather than a carbon copy of the tumor aliquot id)
• [x] Do not truncate RGPU tags
• [x] Ensure all source paths (FQ or BAM) exist and are valid
• [x] Add sex to JSON (cases) to create sex-specific PON for X/Y chromosome events
• [x] Include legacy aliquot ID (esp for TCGA samples, currently only legacy_sample_id included)
• [x] Ensure that final sample-type designation is correct ("TP", "R1", "R2", etc.)

Update 9/10/18: JSON update finished for TCGA, will be used as a model for other cohorts

• [x] TCGA
• [x] Hongkong
• [x] JDG
• [x] HF
• [x] K2
• [x] MDACC

[fpbarthel]

Perhaps the first five characters of flow cell are truly unique? See here

This post from broad uses first five

UPDATE: Confirmed, ~the starting letter and~ four characters following the first are unique. The first character and last four characters define the flow-cell type (eg. Novaseq S2). See here

[Kcjohnson]

Following up on a single point from the list above. Sample TCGA-06-0152 is greylisted in the PCAWG data set. From the PCAWG quality exclusion list:

Greylisted donors will also have consensus variant calls, however in principle these donors are recommended to be used by downstream analyses. Some specific type of analyses may choose to use some of the greylisted donors given that the known QC issue is unlikely affecting the analysis result and the researcher provides clear justification.

There are 10 remaining samples that overlap between PCAWG (2016 data freeze) and GLASS-WG. All of these remaining samples are whitelisted (good quality). 9 GBMs and 1 LGG.