Umccrise tidy

[pdiakumis]

Quite a few enhancements/changes.

• Complete refactor of the dracarys R6 class system. We now have a Wf super class which is the base for Wf_* subclasses.
• Setting up a Wf_umccrise subclass
• S3 function refactor. Now using {paws}/{paws.storage} for everything.
• Adding a umccrise summary reporter

Wf

This class has:

• fields:

  • path: a directory with raw workflow results (can be GDS, S3 or local filesystem)
  • wname: workflow name (e.g. umccrise, sash)
  • filesystem: gds, s3, local
  • regexes: a tibble with file regex and function to parse it

• methods:

  • list_files: List all files under given path.
  • list_files_filter_relevant: List 'dracarys' files under given path
  • download_files: Download files from GDS/S3 to local filesystem (for debugging/exploration)
  • tidy_files: Tidy/process files
  • write: Write tidy tables to database/parquet/tsv/rds etc.

A neat trick it utilises under the hood is eval(parse(text = f), envir = self), where self allows it to use parsers from each individual subclass, which is pretty tricky to get working properly.

Wf_umccrise

• Parses the files within cancer_report_tables/ (signatures for SNV/DBS/Indel, HRD from CHORD/HRDetect, QC summary), conpair/, and the PCGR JSON (for MSI).
• Requires SubjectID/SampleID_tumor in order to more reliably fish out results from the final and work directories (via list_files_filter_relevant).
• Wf_umccrise_download_tidy_write is a neat wrapper that does what it says

S3

• s3_file_presignedurl: uses generate_presigned_url from paws.storage. The client needs to have a v4 signature version (paws.storage::s3(paws.storage::config(signature_version = "s3v4"))).

umccrise summary

• Using Quarto and Reactable.
• Includes:
  • sample metadata
  • summary QC metrics
  • HRD plot (CHORD vs. HRDetect)
  • signatures:
    • table with all of them (for old/new SNVs, DBS, Indels)
    • table with top 2 ranks
    • barplot with top 3 ranks for SNV 2015
  • metadata summary
    • project name/owner
    • source/quality
    • workflow