pirl-unc / hitlist / 24789976505
80%

DEFAULT BRANCH: main
Ran
Jobs 1
Files 21
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: 50.139% (-0.3%) from 50.416%
24789976505

push

github

web-flow 
v1.15.0: generate_species_summary now sources from observations.parquet (#117) (#120)

BREAKING: column names + semantics changed. Before this fix, the
command walked ``pmid_overrides.yaml``'s curated ``ms_samples`` entries
and reported them as data coverage. It undercounted by orders of
magnitude — Mus musculus class I showed ``0 samples`` while the
parquet had 71,755 peptides across 388 PMIDs. Several species present
in the parquet (Rattus, Ailuropoda, Anas, Bubalus, Callithrix, Felis,
Oryctolagus, Pteropus, Xenopus, ...) were missing from the summary
entirely.

The command is now what it looks like it should always have been: a
groupby over observations.parquet by (species, mhc_class).

## Column changes

| Before (yaml-derived)  | After (parquet-derived)     |
|------------------------|------------------------------|
| n_studies              | n_pmids                     |
| n_sample_types         | n_peptides                  |
| n_samples              | n_observations              |

Renaming is deliberate — the old names were misleading
(``n_studies`` didn't mean "studies in the data" but "PMIDs in the
curation YAML"; ``n_samples`` was the sum of a sparsely-populated
field; ``n_sample_types`` was the count of ms_samples rows).

## Sanity

``n_observations >= n_peptides >= n_pmids`` is a structural invariant
(every PMID contributes ≥1 peptide; every peptide contributes ≥1 row).
Asserted in a new test.

## Tests

Replaced the three yaml-era tests with four parquet-era tests:
- ``test_species_summary_columns`` asserts the new column set.
- ``test_species_summary_has_multiple_species`` — unchanged semantics.
- ``test_species_summary_class_filter`` — unchanged semantics.
- ``test_species_summary_covers_non_curated_species`` — asserts
  Mus musculus class I has ``n_pmids > 100`` (real value ~388) and
  ``n_peptides > 10k``. This is the regression the old YAML-based
  implementation silently hid.
- ``test_species_summary_counts_are_coherent`` — the st... (continued)

1623 of 3237 relevant lines covered (50.14%)

0.5 hits per line

Coverage Regressions

Lines Coverage ∆ File
11
83.4
 -2.42% export.py
Jobs
ID Job ID Ran Files Coverage
1 24789976505.1 21
50.14
 GitHub Action Run
Source Files on build 24789976505