pirl-unc / hitlist / 24671512924
80%

DEFAULT BRANCH: main
Ran
Jobs 1
Files 21
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: 44.456% (+0.5%) from 43.981%
24671512924

push

github

web-flow 
v1.11.0: Add bulk proteomics detectability index pilot (#67) (#83)

First increment on #67 — ships a bulk (non-MHC) proteomics index keyed
on cell line, distinct from observations.parquet (MHC MS) and
binding.parquet (in-vitro binding). Purpose: detectability prior per
(sample, source protein) — lets downstream consumers distinguish "not
presented on MHC" from "not detectable in the parent proteome at all."

Source: CCLE proteome (Nusinow et al. 2020, Cell, PMID 31978347). TMT
10-plex whole-cell proteomics across 375 cancer cell lines, log2-
normalized abundances relative to the panel median.

Cell lines filtered to the 7 CCLE lines with substantial hitlist MHC
coverage:

| cell_line    | proteins | hitlist MHC rows |
|--------------|---------:|-----------------:|
| K562         |   10,561 |          47,048  |
| A549         |    9,791 |          23,734  |
| MDA-MB-231   |    9,773 |          98,415  |
| HCT116       |    9,357 |          31,944  |
| THP-1        |    9,147 |         107,018  |
| MCF7         |    9,120 |               ~  |
| Jurkat       |    8,325 |               ~  |

Schema (hitlist/data/bulk_proteomics/ccle_nusinow_2020.csv.gz, 1.6MB):
  cell_line, gene_symbol, uniprot_acc, protein_id,
  abundance_log2_normalized, source, reference

API:
  from hitlist.bulk_proteomics import load_bulk_proteomics, available_cell_lines
  load_bulk_proteomics(cell_line="HCT116", gene_name=["TP53", "KRAS"])

Limitations (documented for follow-up):
- Protein-level, not peptide-level (CCLE doesn't publish per-peptide).
  The peptide-level schema in #67 is left for sources that publish
  per-peptide quantification (e.g. ProteomicsDB).
- HeLa, Raji, 721.221, HEK293, JY — common hitlist hosts — are NOT in
  CCLE. Follow-up will need alternative sources.
- No integration wire into observations.parquet yet — caller-driven join
  via (cell_line, gene_symbol) for now.

Tests: 5 new tests covering cell line list, full load, cell_line filter
(incl. case-insensi... (continued)

1259 of 2832 relevant lines covered (44.46%)

0.44 hits per line

Jobs
ID Job ID Ran Files Coverage
1 24671512924.1 21
44.46
 GitHub Action Run
Source Files on build 24671512924