pirl-unc / hitlist / 34260819756
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: 83.484% (+0.01%) from 83.474%
34260819756

push

github

web-flow 
Honor sample-level override and note, and guard the ms_samples schema (#373) (#443)

generate_ms_samples_table() read most ms_samples keys and silently dropped
three. override sat on 13 samples and note on 3, and nothing anywhere read
either at sample level -- the curator's intent existed only in a field no
code consulted. The exporter iterated the keys it knew and ignored the rest,
so adding a field that no consumer reads looked exactly like adding one that
works.

Export note, classification and reason as their own columns beside the
unchanged legacy notes, and resolve override across four distinct cases
rather than collapsing them to a value: sample (the arm's own claim),
sample_null (a curator considered this arm and decided against one), study
(inherited), none (nobody curated one). sample_null and none produce the
same empty value and mean different things. PMID 34497125 is the shape that
matters -- two cell_line cell-line arms beside a patient-biopsy arm marked
explicitly null.

curation.MS_SAMPLE_FIELDS declares all 22 permitted keys against what reads
each, and loading rejects an undeclared key, so the next silently-ignored
field is an error rather than a discovery. curation.OVERRIDE_VALUES does the
same for the override vocabulary, which previously existed only in
classify_ms_row's branch chain and a YAML header comment -- a misspelled
override fell through to default classification. Study, rule and sample
levels are all validated.

Two limits, stated rather than papered over. The classification flags stay
build-time and PMID/rule-driven: classify_ms_row runs in the scanner, before
any sample attribution exists, so a sample-level override cannot feed them
without recomputing flags in the export and letting it disagree with the raw
index. And origin == "sample" appears on no observation row today, because
all 13 override-bearing samples belong to PMIDs with zero rows in the current
corpus; the sample-export tests cover those cases determinist... (continued)

6849 of 8204 relevant lines covered (83.48%)

0.83 hits per line

Coverage Regressions

Lines Coverage ∆ File
113
90.68
 0.05% export.py
48
94.43
 -0.24% curation.py
Jobs
ID Job ID Ran Files Coverage
1 34260819756.1 30
83.48
 GitHub Action Run
Source Files on build 34260819756