[hyphaltip]

Summary

A population report of A. fumigatus mating type from the sketch store (every public isolate plus the lab collection), shown on a new page of the reads site, plus the two pieces it needs.

Headline: MAT1-1 50.4% per clone (95% CI 48.2–52.5%, 2,077 clones; exact binomial p = 0.76), 51.2% per isolate, across 4,849 isolates (4,736 kept). No country or study departs from 50% at p < 0.01 after clone correction; three near-clonal lineages are skewed (clonal expansion, not a species-wide shift). Sketch and reads-type calls agree 4,614 times and disagree 0 times where both give one idiomorph.

Commits

• 263ef54 ENA FTP fallback (sketch_one.sh): ENA answered HTTPS with 403 for 17 ENA-only runs from the HPCC while FTP served them; the fallback recovered 15 (2 remain ena_partial).
• ad1100b mat_frequency.py: MAT1-1 fraction per isolate and per clone group (ANI 0.9995), by source, study, country and lineage (ANI 0.998), with Wilson 95% CI and an exact binomial test. Leaves out curator exclusions (STORE/exclude.tsv), other species (mean ANI < 0.95) and QC mixtures, with reasons.
• 0087aa8 site report: export_report.py (run on the HPCC) writes datasets/reads/sketch_store/reports/aspergillus_fumigatus/ (population.json 145 KB, runs_public.tsv 438 KB, committed); build_reads_site.py copies it; new web/intro/reads/population.html; the reads overview and the A. fumigatus species report link to it.

Data decisions (curator, 2026-11-09)

• Local Population_Genomics isolates count in every number but are never named: the run table is public runs only, PCoA points carry no identifiers (tested).
• The report is a separate page, not a section of the species report.
• Moose_3-1.1 is excluded (low-quality sequencing; mean ANI 0.749).

Findings worth a look

• 63 isolates filed as A. fumigatus are another species (mean ANI ~0.83): PRJNA1186192 (44, CLSI voriconazole/isavuconazole breakpoints), PRJNA1381729 (6, in vivo breakpoint model), and one run under PRJDB43097 (an A. flavus project). A PopSeq relabelling issue is to follow.
• study_titles.tsv predates those two BioProjects, so they show no title until it is rebuilt from PopSeq.

Test plan

• Full suite: 1955 passed, 18 skipped (pixi run -e test python -m pytest tests -q)
• New tests: test_sketch_store_mat_frequency.py, test_sketch_store_export_report.py (local names never leak), test_build_reads_site_script.py (report copied and indexed; none -> empty list)
• Page in a headless browser (local http.server): 0 console errors, 0 failed requests, all sections filled (17 country groups, 36 studies, 20 lineages, 4,569 PCoA points, 21 screens), no local names, no horizontal scroll at 390 px; overview table and species-page link render

🤖 Generated with Claude Code

[hyphaltip]

Correction to this PR's description (2026-11-10): the 50 runs of PRJNA1186192 (44) and PRJNA1381729 (6) are not another species. Their k-mers match Af293 at ANI 0.991-1.000; they hold a few hundred reads each (likely targeted sequencing filed as WGS), so they are unlike every other sample for lack of data. The isolates of another species are 6 public runs (A. oryzae/flavus-type, A. nidulans, A. fischeri, A. lentulus, 2 unresolved) plus local ones (Moose_3-1.1 is A. nidulans). The MAT frequency does not change (same isolates left out). The page and the exclusion reasons are corrected on branch afum-study-titles; catalog fix proposed in stajichlab/PopSeq#37.