pirl-unc / hitlist / 24996999897
80%

DEFAULT BRANCH: main
Ran
Jobs 1
Files 23
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: 60.667% (+0.2%) from 60.497%
24996999897

push

github

web-flow 
v1.21.0: PMID curation audits (#128, #129, #130, #131, #132) (#160)

* v1.21.0: PMID curation audits (#128, #129, #130, #131, #132)

Resolves a cluster of per-PMID curation issues where the shipped
observations.parquet either misclassified arms or omitted curated
metadata.  Each PMID was audited against its IEDB row payload to
determine the right rule.

#128 — PMID 29789417 (Löffler 2018 CRC).  IEDB exposes the tumor and
matched-normal arms as a single Disease="colonic benign neoplasm"
pool with no structured per-row provenance; the only signal is the
free-text Assay Comments column.  Extend `classify_ms_row` to accept
`assay_comments` and `_matches_condition` to do case-insensitive
substring match on the "Assay Comments" field.  Add three rules
keyed on Assay Comments: CRC-only → cancer_patient, NMC-only →
adjacent, combined "CRC and NMC" → cancer_patient (peptide present
in tumor; NMC-only evidence preserved by NMC-only rows for the same
peptide).  Replaces the over-broad Disease-based rule that was
classifying every row as adjacent.

#129 — PMID 29093164 (Schuster 2017 ovarian carcinoma).  New entry.
IEDB has 1,011 rows split between cancer (Process Type="Occurrence
of cancer") and benign comparator (Process Type="No immunization",
empty Disease) arms; the benign arm would default to healthy_donor
without a rule, so add a Process Type + Disease rule mapping the
comparator arm to adjacent.

#130 — PMID 29331515 (Rozanov 2018 BCC panel).  ~39 IEDB rows carry
non-breast cell names from cross-reference annotation (Assay Comments
"suggesting the elution of a neo-epitope") rather than actual MS
evidence from those non-breast lines.  Document the artifact in the
note and enumerate the 20-line breast cancer panel in ms_samples.
All rows continue to classify as cell_line / cancer.

#131 — PMID 28832583 (Bassani-Sternberg 2017).  IEDB tags every row
as Culture Condition="Cell Line / Clone" (not the EBV-LCL variant),
so the 6 EBV-LCL donor lines were defaulti... (continued)

2545 of 4195 relevant lines covered (60.67%)

0.61 hits per line

Coverage Regressions

Lines Coverage ∆ File
5
86.93
 1.69% curation.py
2
78.92
 0.0% scanner.py
Jobs
ID Job ID Ran Files Coverage
1 24996999897.1 23
60.67
 GitHub Action Run
Source Files on build 24996999897