[ypriverol]

Five datasets show 0 proteins on the PRIDE affinity page. The PRIDE API's noOfProteins comes from this library's proteins-above-lod, which returned an empty list for them. I downloaded the deposited files and found these causes:

Dataset	File(s)	Cause	Fix	Count after fix
PAD000037	Olink Explore media .npx.csv	File has both LODNPX and a count-based LOD (median 1,002); NPX was compared with LOD, so 0 of 1,319 assays passed	Reader prefers LODNPX	868
PAD000024	MDD_rearranged_NPX.npx.csv	Hand-rearranged file with shifted columns (PlateID.1–3, QC.1–3); text in NPX/LOD made the comparison raise	NPX/LOD/LODNPX coerced to numbers; broken rows become NaN	333
PAD000045	v4.0 ADAT	Table-only ADAT with no ^HEADER/^COL_DATA markers; the reader rejected it	Parsed as a ^TABLE_BEGIN body	4,078 (both files combined; v4.0 ADAT via eLOD)
PAD000003, PAD000011, PAD000045 (1.3k ADAT)	SomaScan ADATs	Study samples only (no buffers), so there is no LOD source	Return all measured protein reagents (Type == "Protein") with a warning	6,404 / 6,405 / 1,274

Why fall back to measured proteins for SomaScan: in datasets that do have buffers, nearly every SomaScan assay is above eLOD: 94.9–99.7% per dilution on PAD000021. Olink datasets without any LOD still return none, because many Olink assays really are below LOD.

Also in this PR

• Same LOD as the QC report. get_proteins_above_lod now uses resolve_lod_with_source, moved to processing/lod.py and re-exported from viz.qc.compute. SomaScan therefore uses eLOD rather than an NCLOD computed on buffer samples, matching Use one LOD resolver across the QC report (SomaScan eLOD, FixedLOD shown as active) #30.
• Study samples only. Controls, which mostly sit below LOD, no longer lower the per-assay share. PAD000002 goes from 1,858 to 1,943, and PAD000004 from 149 to 186.

Other PRIDE values will change on regeneration

Several current counts were produced by an older pyprideap build and are stale; for example, main already gives 2,193 for PAD000022 against 1,882 on PRIDE. On local data, the rest move by −8 to +36: PAD000001 1,347→1,361, PAD000006 2,447→2,458, PAD000013 6,384→6,415, PAD000021 6,364→6,400.

Tests

• Updated: SomaScan without buffers now returns measured proteins.
• New tests cover:
  • Olink without any LOD still returns an empty list;
  • Type filter for SomaScan control reagents;
  • controls don't dilute the above-LOD share;
  • LODNPX preference and text coercion;
  • reading a table-only ADAT.
• With CI's versions: ruff clean, mypy clean (45 files), pytest 171 passed.

Deployment

This targets main so it can be part of 1.2.0, which isn't tagged yet. After release, rebuilding pride-affinity-ws regenerates the cached counts (#2 there). PRIDE's noOfProteins is filled from that endpoint, so whatever job writes it needs to run again.

Separately, PAD000024's deposited file has shifted columns, which the curators may want to check with the submitter.

Summary by CodeRabbit

• Bug Fixes
  • Olink data imports now convert measurement and detection-limit values to numbers, preserving invalid entries as missing values. Detection-limit metadata is read from the available source.
  • SomaScan table-only ADAT exports can now be read without section markers.
  • Protein counts above the detection limit now follow platform-specific detection-limit availability. SomaScan counts exclude control reagents, and control samples no longer affect Olink percentages.
  • Blank or placeholder UniProt values are excluded from protein results.

[qodo-code-review]

ⓘ Qodo reviews are paused because the subscription is no longer active. Ask your workspace admin to reactivate the subscription to resume reviews. Manage billing

[coderabbitai]

Note

Currently processing new changes in this PR. This may take a few minutes, please wait...

⚙️ Run configuration

• Configuration used: defaults
• Review profile: CHILL
• Plan: Advanced
• Run ID: a35b8836-bdd8-41e9-9775-a5eed090d18f

📥 Commits

Reviewing files that changed from the base of the PR and between 107fc83 and 41aed2d.

📒 Files selected for processing (5)

• src/pyprideap/io/readers/olink_csv.py
• src/pyprideap/io/readers/somascan_adat.py
• src/pyprideap/processing/lod.py
• src/pyprideap/viz/qc/compute.py
• tests/test_qc_pipeline.py

 ________________________________________________
< I raised Series-B funding to review your code. >
 ------------------------------------------------
  \
   \   (\__/)
       (•ㅅ•)
       / 　 づ

✨ Finishing Touches

📝 Generate docstrings

• Commit to this branch
• Create a new PR

🧪 Generate unit tests (beta)

• Commit to this branch
• Create a new PR

• Autopilot · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}