pirl-unc / tcrsift / 26733055670
84%

DEFAULT BRANCH: main
Ran
Jobs 4
Files 31
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: 79.031% (+0.1%) from 78.886%
26733055670

push

github

web-flow 
2.5.0: closes #118, #119, #120 — allele reasons + cohort novel-allele detection (#121)

* 2.5.0: structured allele-NaN reasons + cohort novel-allele detection

Closes #118, #119, #120.

Allele-call audit (#118 + #120 parts 1-2)
- New per-clone columns:
  * {chain}_allele_called_reason — enum: auto_detected /
    overridden / invalid_override / divergent_contig /
    divergent_at_polymorphic_position / sparse_contig / no_contig /
    no_allele_pool / frame_error.
  * {chain}_observed_constant_aa_start — raw AA translation of the
    contig past the J→C junction (first 15 residues). Surfaces the
    donor's actual residues regardless of whether the picker
    committed to an allele.
  * {chain}_allele_divergence_positions — "pos:expected->observed"
    list for clones where the called allele's canonical AA differs
    from the contig observation.
- Uniformly populate qc_warnings for every no-call path. Pre-2.5
  the sparse-contig case (Mode 2 in the issue) was silent — only
  the divergent-contig case warned, so downstream filters had to
  string-parse the constant_source text.

Tighter tolerance at allele-distinguishing positions (#120 part 4)
- When the picker's best-fit allele disagrees with the contig at
  ANY polymorphic position (a position where packaged alleles
  disagree with each other — e.g. mature TRBC2 pos 9 between *01
  and *03), refuse to commit and emit
  divergent_at_polymorphic_position. The previous behavior could
  silently flip the call to a sibling when overall score won the
  vote but the distinguishing residue failed.

Cohort novel-allele detection (#119 + #120 part 3)
- detect_novel_alleles(df, ...) aggregates per-clone divergences
  by (chain, position, expected_aa, observed_aa), counts clones
  / V-genes / samples, classifies each as novel_allele_candidate
  if pct ≥ min_pct AND n_v_genes ≥ min_v_spread AND n_samples
  ≥ min_samples; else likely_artifact. Defaults from the issue
  (5%, 3, 2) — conservative thresholds validate... (continued)

7176 of 9080 relevant lines covered (79.03%)

3.16 hits per line

Coverage Regressions

Lines Coverage ∆ File
189
52.39
 -0.25% cli.py
110
91.11
 -0.42% assemble.py
Jobs
ID Job ID Ran Files Coverage
1 python-3.11 - 26733055670.1 31
79.02
 GitHub Action Run
2 python-3.9 - 26733055670.2 31
78.98
 GitHub Action Run
3 python-3.12 - 26733055670.3 31
79.02
 GitHub Action Run
4 python-3.10 - 26733055670.4 31
79.02
 GitHub Action Run
Source Files on build 26733055670