27800078173
77%

Ran 19 Jun 2026 01:38AM UTC

Jobs 1

Files 38

Run time 1min

Badge

Embed ▾

Committed 19 Jun 2026 01:37AM UTC coverage: 73.055% (+0.05%) from 73.006%

Build # 27800078173

Build Type

push

github

Committed by

web-flow

Commit Message

Fix NaN crash + NaN cutoff-bypass in tiering/panel; gate RNA-agreement bonus (#136)

Three correctness fixes from the holistic review (regression tests added):

- tiers: int(row.get("rna_somatic_detected_count", 0) or 0) raised on a
  present-but-NaN cell (a left-merge miss vs the HPA RNA consensus). Since this
  runs inside df.apply, one NaN row aborted the whole tiering pass. Add an _int0
  helper (NaN/None -> 0).
- spanning: an unscored peptide (NaN presentation_percentile) slipped through as
  predicted_only because `nan > cutoff` is False. Reject NaN explicitly (matches
  the pd.notna filter in _score_lookup).
- tiers: the RNA "agreement" confidence bonus fired even when protein=SOMATIC
  disagreed with rna=REPRODUCTIVE, inflating a least-safe SOMATIC call to HIGH
  (which passes the selection filter). Gate it on the chosen tissue being
  reproductive. Investigated: 0 bundled CTAs hit this today (the cross-reactive
  -IHC rescue is a separate regen-script override), so this is a zero-impact
  defensive gate against a future regeneration.

Bumps hitlist>=1.43.1 (its data-layer correctness fixes) and 1.20.0 -> 1.20.1.

Coverage Stats

2714 of 3715 relevant lines covered (73.06%)

0.73 hits per line

Coverage Regressions

Lines	Coverage	∆	File
43	77.97	0.23%	tiers.py
6	91.12	0.1%	spanning.py

Jobs

ID	Job ID	Ran	Files	Coverage
1	27800078173.1	19 Jun 2026 01:38AM UTC	38	73.06	GitHub Action Run

pirl-unc / tsarina / 27800078173
77%

README BADGES
x

Markdown

Textile

RDoc

HTML

Rst

Coverage Regressions

Jobs

Source Files on build 27800078173

pirl-unc / tsarina / 27800078173 77%

README BADGES x

Markdown

Textile

RDoc

HTML

Rst

Coverage Regressions

Jobs

Source Files on build 27800078173

pirl-unc / tsarina / 27800078173
77%

README BADGES
x