pirl-unc / tcrsift / 26602724887
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: 77.291% (+0.06%) from 77.229%
26602724887

push

github

web-flow 
2.0.0: uniform J→C junction-residue handling (closes #105) (#106)

Treats α and β chains identically at the J→C splice: read the
junction codon from the contig's first 3 nt past trimmed VDJ,
translate to a per-clone residue, and prepend it to the per-clone
canonical AA. Same code path for both chains.

Pre-2.0 behaviour was asymmetric:
  α: stored TRAC matched UniProt bare mature (starts at I). No
     junction handling — every α chain was 1 aa short of the
     donor's expressed protein at the J→C boundary. The missing
     residue was J-dependent (N/Y/D/H across 118/118 audited B1
     clones).
  β: a synthetic E was prepended to HUMAN_TRBC1_AA / HUMAN_TRBC2_AA
     at FASTA-write time, so the assembled β chain had the E at
     the junction. The hardcoding worked for β because the
     J→C junction codon is universally GAG (E) across all 118
     audited B1 clones — but it baked the universal-E case into
     the canonical AA instead of treating it as data.

2.0 behaviour: both chains read the junction codon from the
contig's first codon past trimmed VDJ. The canonical sequences
in the packaged FASTA are now BARE MATURE for all three genes,
matching UniProt directly. The junction residue is prepended
per-clone in ``_add_constant_regions``. NT bytes for the
junction codon come from the donor's actual contig (preserving
donor fidelity at the splice boundary).

Breaking changes:
  - ``HUMAN_TRBC1_AA`` and ``HUMAN_TRBC2_AA`` no longer have a
    synthetic ``E`` at index 0. Direct consumers of these
    constants see different values (176/178 chars instead of
    177/179, content shifted by 1 from idx 0). The
    ``CONSTANT_REGION_ENDINGS`` table is unchanged — endings are
    independent of the prepend.
  - β chains assembled WITHOUT contigs no longer get the
    automatic E prepend. The assembled mature β is 1 aa shorter
    than what the donor would express, with a QC warning to that
    effect. Run with ``--contigs-dir`` for the biologically
    cor... (continued)

6562 of 8490 relevant lines covered (77.29%)

3.09 hits per line

Coverage Regressions

Lines Coverage ∆ File
62
90.63
 0.25% assemble.py
15
0.0
 0.0% _regen_canonical_constants.py
Jobs
ID Job ID Ran Files Coverage
1 python-3.10 - 26602724887.1 31
77.28
 GitHub Action Run
2 python-3.12 - 26602724887.2 31
77.28
 GitHub Action Run
3 python-3.11 - 26602724887.3 31
77.28
 GitHub Action Run
4 python-3.9 - 26602724887.4 31
77.25
 GitHub Action Run
Source Files on build 26602724887