pirl-unc / hitlist / 25282349993
80%

DEFAULT BRANCH: main
Ran
Jobs 1
Files 27
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: 71.315% (-0.05%) from 71.36%
25282349993

push

github

web-flow 
v1.30.31: Resolve C1R-cell-line alleles for 4 PMIDs (#207) (#220)

Closes #207 — C1R curation gap where ~4,121 IEDB rows on the
HLA-class-I-deficient C1R B-LCL transfectant carried only the
class-only "HLA class I" placeholder. Paper Methods read on each
of the 4 PMIDs:

- 27846572 (Liepe 2016): parental C1R, endogenous low expression
  HLA-B*35:03 + HLA-Cw*04:01. 3,206 rows on cell_name="C1R cells-
  B cell" now correctly resolve to the C1R parental sample.
- 27920218 (Alpizar 2017): the 515 unattributed rows are pooled
  across the same C1R-B*40:02 and C1R-B*39:01 transfectants used
  elsewhere in the paper. Per-row APC text says "C1R-B*40" or
  "C1R-B*39" — the resolver's allele-substring tie-break splits
  408 to B*40:02, 99 to B*39:01, 8 to a new pooled entry that
  owns the rows whose APC mentions both transfectants.
- 30778243 (Koutsakos 2019, flu A/B/C cross-reactivity): NEW
  YAML entry. C1R parental (w6/32) + C1R.A*02:01 transfectant
  (BB7.2) for class I, plus LCL-endogenous DR/DQ/DP for class II
  (LB3.1 + SPV-L3 + B721 cocktail). Class-II-on-C1R rows are
  legitimate and now allocate.
- 34006841 (Hensen/Habel 2021, HLA-A*24:02 in Indigenous pops):
  NEW YAML entry. C1R.A24 transfectant (class I) + endogenous
  HLA-DR12 / DPB1*04:01,04:02 / DQB1*03:01 (DQ7) for class II.
  All 179 rows now resolve.

Also tightens the v1.30.30 class-pool tie-break:

- Allele-substring matching: the previous min_len=3 token cutoff
  silently dropped allele-locus tokens like "B*40" / "B*39" that
  IEDB writes into APC narrative. Score 1 per substring hit on
  each candidate's curated mhc field. This is what splits the
  Alpizar 515 rows.
- Two-pass scoring: pass 1 uses cell_name + source_tissue only;
  pass 2 falls back to APC + AC ONLY when cell_name was constant
  across the group. Without the cell_name-varies guard, AC
  narrative mentions of generic descriptors (e.g. Liepe's "GR
  lymphoblastoid cell line") were cherry-picking the wrong sample
  via the C... (continued)

3851 of 5400 relevant lines covered (71.31%)

0.71 hits per line

Coverage Regressions

Lines Coverage ∆ File
119
81.52
 -0.43% export.py
Jobs
ID Job ID Ran Files Coverage
1 25282349993.1 27
71.31
 GitHub Action Run
Source Files on build 25282349993