pirl-unc / hitlist / 25806660708
76%

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: 75.86% (+0.3%) from 75.596%
25806660708

push

github

web-flow 
v1.30.51: group pmhc output by inferred MHC species (closes #256) (#257)

* v1.30.51: group pmhc output by inferred MHC species (closes #256)

When a `hitlist pmhc --gene FOO --predictor netmhcpan` query returns
evidence across multiple MHC species (HLA + DLA + H-2 because
netMHCpan is pan-allele and the corpus has cross-species data), the
rows interleaved by gene without any visual cue that a row was on a
non-human MHC.  Easy to misread canine PRAME as human evidence.

Two changes:

1. Result DataFrame now carries an `mhc_species` column derived from
   `mhc_allele` via `hitlist.curation.classify_mhc_species` (mhcgnomes
   with a prefix regex fallback).  CSV / JSON consumers get it too.

2. format_table() inserts `=== species: X ===` outer headers when the
   result has more than one species.  Single-species results (the
   typical human-only case) skip the header so output stays compact.
   Order: human → mouse/rat → alphabetical → "other" sinks to bottom.

5 new tests (32 total in test_pmhc_query.py): column attachment,
species inference for HLA / DLA / H-2 / empty / garbage allele
strings, sort-key order, multi-species sectioning, single-species
no-section.

* fix mhc_species drop on predictor path (closes #256 review bug)

Original PR derived mhc_species BEFORE _attach_predictions, but
_consolidate_after_narrowing groups by a fixed column list and
silently drops anything not in it.  Result: hitlist pmhc --predictor
netmhcpan returned a frame with no mhc_species column → format_table
never sectioned by species → the whole feature was a no-op for the
exact use case (`--gene PRAME --predictor netmhcpan`) that motivated #256.

Fix: derive mhc_species AFTER _attach_predictions.  Bonus: it now
reflects the post-narrowing single-allele string instead of the
multi-allele pre-narrowing typing string (which would parse to
"other" via classify_mhc_species).

Regression test verified to fail without the prod fix and pass with it.

* pmhc_query: use authorita... (continued)

4497 of 5928 relevant lines covered (75.86%)

0.76 hits per line

Coverage Regressions

Lines Coverage ∆ File
24
91.82
 2.61% pmhc_query.py
Jobs
ID Job ID Ran Files Coverage
1 25806660708.1 28
75.86
 GitHub Action Run
Source Files on build 25806660708