pirl-unc / tcrsift / 26575381279
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: 77.011% (-0.4%) from 77.375%
26575381279

push

github

web-flow 
1.2.0: fix #100 — canonical TCR constants sourced from pyensembl FASTA (#102)

Previous hardcoded HUMAN_TRAC_AA / HUMAN_TRBC1_AA / HUMAN_TRBC2_AA in
assemble.py drifted from their cited UniProt source at 7 positions
across the three genes (#100):

  TRAC  : pos 47   C → T (should be T per UniProt P01848)
  TRBC1 : pos 135  E → V (swapped with TRBC2 pre-#100)
  TRBC2 : pos 3-4  NK → KN (the JOVI.1-distinguishing K/N swap;
                              both chains read EDLNKVF pre-#100)
          pos 9    E → K  (FPPE → FPPK)
          pos 56   C → S  (SGVCT → SGVST)
          pos 135  V → E  (swapped with TRBC1)

#100's provenance trace found the cited UniProt accession for TRBC2
(A0A075B6Y0) was wrong — it's actually TRAJ49 — so the original PR #69
intent of "match UniProt" never had a verifiable check. Sequences
likely came from "prototype code from notebooks" (commit 2f438c5) with
hand-typing errors.

Fix:
  - tcrsift/refseqs/canonical_constants.fasta: new packaged data file
    generated from pyensembl GRCh38 release 110 protein_sequence for
    canonical transcripts ENST00000611116 / ENST00000633705 /
    ENST00000466254. Cross-checked against UniProt P01848 / P01850 /
    A0A5B9 — matches exactly after stripping pyensembl's leading X
    splice-boundary placeholder and applying tcrsift's prepend-E
    convention for β chains.
  - tcrsift/assemble.py: HUMAN_TRAC_AA, HUMAN_TRBC1_AA, HUMAN_TRBC2_AA
    now loaded from the packaged FASTA at module import time via
    importlib.resources. Module-level invariant asserts each sequence
    ends with the canonical C-terminus from CONSTANT_REGION_ENDINGS.
  - tcrsift/_regen_canonical_constants.py: helper module to regenerate
    the FASTA from pyensembl. Run as
        python -m tcrsift._regen_canonical_constants
  - tests/test_assemble_constants.py:
      + test_diagnostic_positions_match_uniprot — pins each of the 7
        residues that drifted, so future hand-edits fail loudly.
      + test_jovi1_disti... (continued)

6472 of 8404 relevant lines covered (77.01%)

3.08 hits per line

Coverage Regressions

Lines Coverage ∆ File
73
89.3
 0.18% assemble.py
Jobs
ID Job ID Ran Files Coverage
1 python-3.11 - 26575381279.1 31
77.0
 GitHub Action Run
2 python-3.12 - 26575381279.2 31
77.0
 GitHub Action Run
3 python-3.10 - 26575381279.3 31
77.0
 GitHub Action Run
4 python-3.9 - 26575381279.4 31
76.97
 GitHub Action Run
Source Files on build 26575381279