Repository navigation
Reads: species where presence of an idiomorph cannot be typed (silent cassettes, homothallic, switching) and what the gate should say #195
Description
Problem
Presence typing assumes that a cell carries the sequence of one idiomorph. It fails, or reads both in every strain whatever the active MAT, where the genome holds silent copies or both genes in one locus:
- Schizosaccharomyces pombe: mat1 plus the silent mat2-P / mat3-M cassettes. Confirmed in wave 2: the lineage-1 representative (GCA_006386395.1, SNRJ01000005.1:2442048-2478409, 36 kb) is the same locus for P and M and the two cores are 14,111 bp of identical sequence, so the panel has no unique k-mers (
panel_no_unique_kmers, waspanel_L1_error). - Sclerotinia sclerotiorum: homothallic, both MAT genes in one locus;
panel_homothallic.pyexists from the pilot. - Ceratocystis fimbriata: same hold; cause not checked (unidirectional switching is published for Ceratocystis and is a hypothesis here; cite or test).
- Saccharomycotina with HML/HMR silent cassettes (Saccharomyces, Kluyveromyces, Lachancea, Zygosaccharomyces): the same structural problem is expected, and Komagataella phaffii (50 of 50 runs
both, in line with its two-locus genome) shows it. This is not yet checked per species.
The gate (wave_driver.py, homothallic_both=0.5) assigns PROVISIONAL_HOMOTHALLIC on a share of both alone. Homothallism, silent cassettes, diploid heterozygosity, mixed or contaminated libraries and non-specific panels all look the same, so the status can publish a structural artefact as biology.
Proposed
- Decide the wording on the site for these species, for example "not typable from reads by presence: both idiomorphs are expected in every cell" (curator to approve the exact text), and stop presenting their single-call shares as allele frequencies.
- Separate the reasons in the gate output: homothallic (both genes in one locus), silent-cassette/switching, heterozygous diploid. The page text should name which.
- Active-MAT typing needs junction k-mers (MAT Y/Z plus the unique flank at the MAT locus, not the HML/HMR flanks) or read-depth ratios at the MAT junction. Prototype on S. pombe (mat1 state) and one Saccharomyces species.
- C. fimbriata: check the cause on the assemblies.
Tests / acceptance
- S. sclerotiorum positive control: at least 95% of runs
bothand 0 single calls (the current gate would pass at 50%). - For any junction panel: simulated mixes of a and alpha reads at known ratios; minor idiomorph detected at 10% or more share in 3 of 3 replicates (same style as the Fola test in Landing page: show the Fola and A. fumigatus read-typing results like Coccidioides (charts, per-strain tables) #172).
Related: Raw-read MAT typing of public fungal population datasets (survey, panels, pipeline) #173, Reads: yeast panels (Saccharomyces sensu stricto, Candida albicans, Zygosaccharomyces, Lachancea and others) withheld in wave 2 #196.
Activity
hyphaltip commentedon Nov 8, 2026
Status (2026-11-07, reads session; work in PR #283 and #282):
Done:
wave_driver.ZYGOSITY(Candida albicans, Pichia kudriavzevii, curator rulings 2026-11-06/07):bothcounts as a call (MTL a/alpha) and agrees with an assembly's single idiomorph; wordsPASS_ZYGOSITY/PROVISIONAL_ZYGOSITY, shown on the site as "(MTL zygosity)". The homothallic shortcut no longer applies to them. This covers proposed point 2 for one of the three reasons.wave_driver.RELEASED(S. paradoxus, S. eubayanus, S. pastorianus, S. mikatae, Ogataea haglerorum) getUNVALIDATED:released_cassettes. With up to 300 runs each the panels give essentially no calls: S. paradoxus 0 of 299 typed, S. eubayanus 2 of 301, S. pastorianus 11 of 91, O. haglerorum 0 of 23. That confirms presence typing does not work for this group.Not done (the curator deferred the new panel design on 2026-11-07):
Leaving open for the panel-design work.
hyphaltip commentedon Nov 9, 2026
Curator ruling 2026-11-09 (design note in PR #321): this issue closes for v0.9.0 when these merge.
wave_driver.py:homothallic_locus,silent_cassettes,zygosity(exists),cause_unchecked. These replace the 50%-both PROVISIONAL_HOMOTHALLIC shortcut for these species.homothallic_locusorsilent_cassettes.silent_cassettes, and re-check H. anomala and M. pulcherrima.cause_uncheckedkept with what was found).homothallic_locus. The positive-control acceptance (at least 95%both) moves with the panel work.Moved to v0.10.0: the copy-number ratio test (#322) and the linked-pair junction panel (#323).
Notes: junction-panel and cannot-be-typed design for #195/#196 (propo…
Gate causes (#195/#196): causes.tsv, NOT_TYPABLE:homothallic_locus|si…
Reads gate causes for v0.9.0 (#195, #196): not typable, zygosity, unc…
hyphaltip commentedon Nov 12, 2026
Curator decisions 2026-11-12 on the open questions of
docs/notes/2026-11-09_junction-panel-design.md(#321), for whoever does the reads-page work (theretest-downloads-genus-panelssession was not running to receive this directly):homothallic_locus"Homothallic: both MAT genes in one locus in every genome.";silent_cassettesthe text in the note;cause_unchecked"Withheld: cause not yet checked.";zygosityunchanged; technical reasons withheld with the reason. S. uvarum stays UNVALIDATED; S. sclerotiorum and C. fimbriata keep their holds.silent_cassettes. H. anomala and M. pulcherrima: check their reads first, decide in v0.10.Recommendations and these decisions are also in PR #340.
hyphaltip commentedon Nov 12, 2026
Correction to my comment above (2026-11-12): the Q3 relabel is already on main.
datasets/reads/sra_survey/pilot/wave2/gates_final.tsvgives K. marxianus, K. phaffii and S. kudriavzeviiNOT_TYPABLE:silent_cassettes, andcauses.tsvrecords each as "was PROVISIONAL_HOMOTHALLIC". Nothing to relabel. The remaining question is wording:CAUSE_TEXTinscripts/build_reads_site.py(curator ruling 2026-11-09) wordshomothallicandcause_uncheckeddifferently from the draft table in section 2 of the design note that the curator approved today. Asked the curator which to keep.hyphaltip commentedon Nov 12, 2026
Curator decision 2026-11-12: keep the wording already on the site (
CAUSE_TEXT, ruling of 2026-11-09). The draft table in section 2 of the design note is superseded; no change tobuild_reads_site.pyor to #337. PR #340 is updated.hyphaltip commentedon Nov 12, 2026
Released in v0.9.0.
PR #325 implements the checklist, following the ruling on the design note (#321):
datasets/reads/sra_survey/pilot/causes.tsvholds one cause per species (42 species), with the evidence incause_evidence.tsv.NOT_TYPABLE:homothallic,NOT_TYPABLE:silent_cassettes, zygosity, andUNVALIDATED:cause_unchecked. The 50%-bothshortcut no longer givesPROVISIONAL_HOMOTHALLIC.NOT_TYPABLE:silent_cassettes.CAUSE_TEXTfrom the ruling of 2026-11-09. The curator kept it on 2026-11-12.The junction-panel and ratio-test work continues in #323 and #322 (v0.10.0). Closing.