Skip to content

fix(cluster): robust benchmark/uniprot downloads (User-Agent + optional cache)#231

Merged
mat10d merged 1 commit into
mainfrom
fix-uniprot-mouse
Jul 10, 2026
Merged

fix(cluster): robust benchmark/uniprot downloads (User-Agent + optional cache)#231
mat10d merged 1 commit into
mainfrom
fix-uniprot-mouse

Conversation

@mat10d

@mat10d mat10d commented Jul 10, 2026

Copy link
Copy Markdown
Collaborator

What

Robustness fixes for the benchmark/annotation downloads in lib/cluster/scrape_benchmarks.py (ported from zarr3):

  • User-Agent header on the UniProt / STRING / CORUM / MSigDB REST calls — headerless requests are increasingly rejected by these services (a real cause of silent download failures).
  • Optional cache_path on get_uniprot_data — if the TSV exists it's loaded without a network call, and a fresh fetch is written there for reuse.

Supports #229 (annotation downloads failing). get_uniprot_data already accepts species_id; the analysis-side companion PR exposes it in the configure notebook so non-human screens work.

ruff + compile clean; defaults unchanged (cache_path=None ⇒ always fetch).

🤖 Generated with Claude Code

…al cache)

Add a User-Agent header to the UniProt/STRING/CORUM/MSigDB REST calls (headerless
requests are increasingly rejected) and an optional cache_path to get_uniprot_data
so repeat runs load from disk instead of re-querying. Ported from zarr3. Helps
issue #229 (annotation downloads failing).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
@mat10d mat10d merged commit d976c0d into main Jul 10, 2026
3 checks passed
@mat10d mat10d deleted the fix-uniprot-mouse branch July 10, 2026 19:04
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant