pirl-unc / hitlist / 33023104588
81%

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: 80.852% (+0.08%) from 80.772%
33023104588

push

github

web-flow 
v1.48.0: derive MHC class/alleles from mhcgnomes; fix non-classical pooling (#363) (#369)

Replaces bespoke allele/class regexes with mhcgnomes across the parsing
paths, and fixes the two defects that fell out of them.

New in curation.py, all mhcgnomes-backed:
  mhc_class_of              I / II / non-classical, derived not matched
  normalize_mhc_class_token bridges "non classical" vs "non-classical"
  is_class_only_token       MhcClass detection, any species' notation
  expand_allele_components  Pair splitting instead of str.split("/")
  extract_allele_tokens     species-agnostic token extraction

mhc_class_of reproduces the curated class on 2,922,227 observation rows
and differs on 72 — Caja-E and Mamu-E*02:11 are the marmoset/rhesus
MHC-E genes, curated as classical but non-classical in fact.

samples could never be pool candidates, and the YAML's "non-classical"
could never equal the IEDB export's "non classical". Both fixed; the
pool now iterates the classes actually present. --mhc-class I still
does not sweep up HLA-E rows, since non-classical keeps its own token.

Mouse-allele drop: _parse_sample_mhc_field used an HLA digit-syntax
regex, so "H-2Kb H-2Db", "H-2Q1 H-2Q2" and "Patr-AL" all parsed to the
empty set and per-peptide attribution could never narrow a non-human
sample's candidate alleles.

extract_allele_tokens accepts only Allele/Gene/Pair: mhcgnomes resolves
"n/a" to the rat haplotype RT1-n/A, so accepting anything that merely
parses would inject junk from free-text fields.

Closes #363

Claude-Session: https://claude.ai/code/session_014cBpazLoFvftLSLavT7T9X

6051 of 7484 relevant lines covered (80.85%)

0.81 hits per line

Coverage Regressions

Lines Coverage ∆ File
104
85.66
 -0.06% export.py
40
92.11
 0.18% curation.py
Jobs
ID Job ID Ran Files Coverage
1 33023104588.1 30
80.85
 GitHub Action Run
Source Files on build 33023104588