pirl-unc / hitlist / 34184009546
84%

DEFAULT BRANCH: main
Ran
Jobs 1
Files 30
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: 83.474% (+0.02%) from 83.456%
34184009546

push

github

web-flow 
Conserve every curated ms_samples record from YAML to export (#438, #437) (#439)

* Conserve every curated ms_samples record from YAML to export (#438, #437)

Two independent, silent leaks removed curated sample records between
pmid_overrides.yaml and every consumer.

load_pmid_overrides() built its mapping with a dict comprehension keyed
by PMID, so a second entry for an already-curated PMID replaced the
first — nothing raised, nothing warned. #307 added source-organism
curation as new blocks for 33460454 and 28188227, which displaced the
study blocks holding five sample records (three Tasmanian devil, two
HLA-B27 transgenic rat). Consolidate both pairs into single entries and
reject duplicate identifiers at load. Consolidation keeps the exact
species strings the scanner already read, so no provenance fill moves,
and corrects two study labels that named the wrong first author
(Gastaldello 2021, not "Owen 2021"; Barnea 2017, not
"Alvarez-Navarro 2018"), both verified against PubMed.

generate_ms_samples_table() dropped every n_samples: 0 record — exactly
the explicit profiled: false curation the profiled column exists to
distinguish from uncurated. Export them as metadata rows instead, and
exclude them from observation attribution in the join, so restoring
them cannot manufacture evidence. That guard is load-bearing:
attribution path 3c matches attributed_sample_label against
sample_label with no allele involved and overrides every heuristic
above it.

Verified against the full local corpus: sample export 739 -> 748 rows
(the nine records described above, none removed, no field changed on
any pre-existing row); observation rows 4,439,321 unchanged, with pmid,
peptide, mhc_restriction, source_organism and species identical
row-for-row. The only observation change is metadata gained: 33,959
Tasmanian-devil rows now discriminate into their three curated arms.
28188227 stays unattributed, correctly — both arms carry the same
imprecise mhc: HLA-B*27.

READ... (continued)

6829 of 8181 relevant lines covered (83.47%)

0.83 hits per line

Coverage Regressions

Lines Coverage ∆ File
99
90.63
 0.03% export.py
46
94.67
 0.03% curation.py
Jobs
ID Job ID Ran Files Coverage
1 34184009546.1 30
83.47
 GitHub Action Run
Source Files on build 34184009546