pirl-unc / hitlist / 25562739614
75%

DEFAULT BRANCH: main
Ran
Jobs 1
Files 28
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: 74.185% (+0.1%) from 74.071%
25562739614

push

github

web-flow 
v1.30.42: fix categorical fillna in export + cartesian-product hang in pmhc_query (#235)

Two related categorical-handling bugs surfaced after the post-#137
categorical compression of observation-frame string columns:

1. **export.py — `_fillna_safe_for_categoricals` helper.** Three
   `loc[mask, cols].fillna("")` sites in the per-PMID
   discriminator-scoring path of `generate_observations_table` raised
   `TypeError` when the slice contained a categorical column whose
   category set didn't include `""`.  New helper casts categoricals to
   `StringDtype` first; same `fillna("")` semantics, no crash.

2. **export.py — `generate_species_summary`.**  Filtering
   `load_observations` by `mhc_class="I"` left the categorical's
   `categories` set carrying `"II"` and `"non classical"`.  Without
   `observed=True`, `groupby(dropna=False)` then emitted phantom
   zero-row bins for every dropped level per species.  Added
   `observed=True`.

3. **pmhc_query.py — `observed=True` on all 5 groupbys.**  The
   aggregation `groupby` over six keys (gene_name, gene_id,
   mhc_restriction, best_guess_allele, peptide, mhc_class) was the
   cartesian product of all category levels — billions of empty groups
   on a 4.4M-row frame.  This both emitted the pandas FutureWarning at
   line 261 and caused `hitlist pmhc --gene <X>` to hang indefinitely
   after loading observations.  Same fix applied prophylactically to
   the four iteration-style groupbys at lines 568/600/607/611, which
   also walk potentially-categorical keys.

Tests: two regressions in test_export.py for the helper and the
species-summary class filter.  All 671 tests pass.

4118 of 5551 relevant lines covered (74.18%)

0.74 hits per line

Coverage Regressions

Lines Coverage ∆ File
138
80.61
 0.71% export.py
14
88.36
 0.0% pmhc_query.py
Jobs
ID Job ID Ran Files Coverage
1 25562739614.1 28
74.18
 GitHub Action Run
Source Files on build 25562739614