| Age | Commit message (Collapse) | Author |
|
Track marker names per chromosome during Stage 03 QC and report an error for
duplicates within the same chromosome. Marker names may still be reused on
different chromosomes.
Co-authored-by: pi coding agent <pi@pi.dev>
|
|
Add `uploader.genotypes.normalise`, which writes a canonical untransposed
matrix (`<stem>__work<suffix>`) next to the original, dropping comments and
`@` directives and leaving the original untouched. Ragged or empty matrices
raise `NormalisationError`, and `qc.check_file` now handles only the normal
orientation.
Stage 02 now puts `stage02`, `genotype_collector_job_id`, `file` and `config`
in the QC job's metadata at launch, and uses `jobs_by_metadata` to reuse an
existing QC job. The completed collector job's metadata is no longer modified.
The Stage 03 worker normalises and QCs the working file from its own metadata.
Change the placeholder marker ordering to chromosome, then Mb, then cM.
Co-authored-by: pi coding agent <pi@pi.dev>
|
|
- Store the genotype file path and collector config in the collector job
metadata so later stages can re-read the file.
- Seed `dot_geno_fields.unk` into the Stage 02 missing-value symbols default.
- Add `uploader.genotypes.qc` with value checks (empty and unknown symbols)
and a placeholder marker-ordering helper (chromosome, cM, Mb) that warns.
- Add the `uploader.scripts.genotypes.qc` worker that reads the collector job,
runs the checks and exits non-zero when errors are found.
- On Stage 02 save, schedule the QC job and record its id for idempotency.
|