[hyphaltip]

What

On each reads species page, study accessions now link to NCBI BioProject, and the link's hover text is the project title.

• Links: PRJNA…/PRJEB…/PRJDB… link to https://www.ncbi.nlm.nih.gov/bioproject/<acc>. Bare SRP/ERP/DRP accessions link to a BioProject search instead.
• Where: the study column of the runs table, and the "By study" rows of the counts table. The "All studies pooled" row stays plain text.
• Hover: for example, PRJNA624742 shows "Genomic determinants of host specialization in the polyphagous fungal pathogen Botrytis cinerea".
• Search: the runs search box also matches project titles.
• Wording: the outdated "up to 50 per species, at most 4 per study" text on the reads index now reads "up to 700 per species".

Where the titles come from

• New script: datasets/reads/sra_survey/pilot/study_titles.py collects the studies in the committed wave tables. It takes each study_title from the PopSeq kingdom run table (cache/fungi_runs_ena_2026-11-06.tsv.gz), so PopSeq stays the metadata source.
• New table: study_titles.tsv, committed, holds 3,912 studies, all with a title (3,663 PRJNA, 174 PRJEB, 75 PRJDB).
• Builder: scripts/build_reads_site.py adds study_titles to each species JSON, limited to that species' own studies. If the TSV is missing, the map is empty.
• Docs: documented in CONTRACT.md.

Checks

• Tests: 2 new tests in tests/test_build_reads_site_script.py.
• Full suite on the HPCC at ad0c89c: 1951 passed, 6 skipped.
• Headless browser, Botrytis cinerea page:
  • 338 of 338 run rows linked, and 14 of 14 study rows in the counts table.
  • PRJNA624742 has the correct href and its title on hover.
  • No console errors.

The live site changes only after this is merged and then deployed, either with a version tag or with a manual "Deploy browser page" run with build_only unticked.

🤖 Generated with Claude Code