25073366557
80%

Ran 28 Apr 2026 07:32PM UTC

Jobs 1

Files 25

Run time 1min

Badge

Embed ▾

Committed 28 Apr 2026 07:30PM UTC coverage: 68.99% (+0.3%) from 68.718%

Build # 25073366557

Build Type

push

github

Committed by

web-flow

Commit Message

v1.30.1: pmhc binder classification — affinity + percentile, take strongest (#184)

Replace the old binary "binder yes/no" output with a 4-tier classification
that uses both signals from the predictor and reports the strongest call:

  Affinity (IC50, nM):       strong ≤ 100,  medium ≤ 500,  weak ≤ 2000,  non-binder > 2000
  Percentile (rank, %):      strong ≤ 0.5,  medium ≤ 1.0,  weak ≤ 2.0,    non-binder > 2.0

Why both: predictors disagree more about absolute IC50 than about per-allele
rank; a peptide that scores "strong" by percentile but only "weak" by IC50
is reported as strong (the union of evidence rather than the intersection).

The grouped text format now shows pct_rank as its own column alongside
affinity_nM and the binder tier.

Coverage Stats

3266 of 4734 relevant lines covered (68.99%)

0.69 hits per line

Coverage Regressions

Lines	Coverage	∆	File
7	83.72	6.39%	pmhc_query.py

Jobs

ID	Job ID	Ran	Files	Coverage
1	25073366557.1	28 Apr 2026 07:32PM UTC	25	68.99	GitHub Action Run

pirl-unc / hitlist / 25073366557
80%

README BADGES
x

Markdown

Textile

RDoc

HTML

Rst

Coverage Regressions

Jobs

Source Files on build 25073366557

pirl-unc / hitlist / 25073366557 80%

README BADGES x

Markdown

Textile

RDoc

HTML

Rst

Coverage Regressions

Jobs

Source Files on build 25073366557

pirl-unc / hitlist / 25073366557
80%

README BADGES
x