pirl-unc / hitlist / 24932536346
80%

DEFAULT BRANCH: main
Ran
Jobs 1
Files 23
Run time 1min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst
coverage: 59.567% (+0.2%) from 59.318%
24932536346

push

github

web-flow 
v1.19.1: Line-expression cache fingerprints (#150) (#157)

Issue #150 listed several integration blockers for the line-expression
WIP.  The first three (package-data globs, GM12878 gene_name
population, DepMap ModelID harmonization) shipped in v1.16.0 / v1.17.0.
The remaining one is cache fingerprinting: build cache fingerprints
omitted line-expression curation files and registered DepMap inputs,
so an anchor-YAML edit or a freshly-registered DepMap CSV would not
invalidate the cache and the line-expression parquet would silently
go stale.

- hitlist/builder.py:
  - New _stat_fingerprint helper consolidates the per-file
    {path, size, mtime} dict that was duplicated in two places.
  - _source_fingerprints now stamps:
    - data/line_expression_anchors.yaml
    - data/line_expression/sources.yaml
    - every packaged data/line_expression/*.csv.gz
    - every registered depmap_rna / depmap_rna_transcript /
      depmap_models input from the downloads manifest
    so any curation change OR a re-registered DepMap export forces
    the next 'hitlist data build' to rebuild line_expression.parquet.
  - _parquet_fingerprints adds line_expression.parquet to the
    fingerprint set so a swapped-in parquet invalidates the cache too.
  - _cache_is_valid now also requires line_expression.parquet to be
    present (alongside observations / binding / bulk_proteomics).

- tests/test_builder.py (6 new):
  - line_expression_anchors.yaml fingerprinted by _source_fingerprints.
  - line_expression/sources.yaml fingerprinted.
  - Packaged GM12878 CSV fingerprinted (any *.csv.gz under the dir).
  - Registered DepMap input flows into the source fingerprint dict.
  - line_expression appears in _parquet_fingerprints.
  - Missing line_expression.parquet alone invalidates the cache.
  Updated the two existing cache-validity tests to also seed a
  line_expression.parquet so they keep passing.

Version: 1.19.0 → 1.19.1 (patch — internal cache fix, no API change).

2478 of 4160 relevant lines covered (59.57%)

0.6 hits per line

Coverage Regressions

Lines Coverage ∆ File
172
60.35
 1.52% builder.py
Jobs
ID Job ID Ran Files Coverage
1 24932536346.1 23
59.57
 GitHub Action Run
Source Files on build 24932536346