Repository navigation
Reads: copy-number ratio test for active MAT in cassette yeasts and MTL zygosity #322
Description
Split from #195 and #196 by the curator ruling of 2026-11-09 on the design note docs/notes/2026-11-09_junction-panel-design.md (PR #321), design A.
What
Silent cassettes give every haploid both idiomorph sequences, so presence typing cannot say which one is active. Their copy number can. With the standard cassettes (HMLalpha, HMRa), the depth ratio of Ya-unique to Yalpha-unique k-mers (depth_1, depth_2) is:
| State | Ya:Yalpha |
|---|---|
| haploid MATa | 2:1 |
| haploid MATalpha | 1:2 |
| diploid a/alpha | 1:1 |
S. pombe is the same with P and M. C. albicans has no cassettes, so the ratio there is a straight zygosity test: 1:1 is a/alpha, and one side near zero is homozygous after loss of heterozygosity.
Steps (costs from the note)
- Opt-in policy in
reads-typeand the gate, defaultoff, documented indocs/detect-options.md. Classes:a,alpha,a/alpha,switching/mixed,indeterminate. About 1 day with tests. - C. albicans and P. kudriavzevii first. No new runs are needed, because
depth_1anddepth_2are already in the wave tables. Report the share of homozygous strains, and check against a published MTL zygosity table (curator to name it). - Rebuild Saccharomyces sensu stricto and S. pombe panels from curated tier A/A1 records. For S. pombe, build from the P and M cassettes, not the 36 kb three-locus representative. About 1 day.
- Simulated-mix acceptance per rebuilt panel. Mixes at 0, 10, 25, 50, 75, 90 and 100%, 3 replicates each, about 60 short jobs per panel.
- Pass if the minor idiomorph is detected at a 10% share or more in 3 of 3 replicates.
- Pass if haploid a, haploid alpha and diploid a/alpha are classified correctly at 10x or more.
- Controls: S. cerevisiae BY4741 (a), BY4742 (alpha), BY4743 (a/alpha) and S288C; S. pombe 972 h−, 975 h+ and 968 h90, where h90 is the expected
switching/mixedresult. - Re-type the cassette species. About 1,000 Saccharomyces runs; download-bound, about a day on the cluster.
Known limits
The ratio breaks on non-standard cassettes (such as HMLa), aneuploidy, mixed libraries, and switching cultures. Those should come out as switching/mixed or indeterminate, never as a call.
If the controls fail, see #323 (linked-pair junction panel).
Related: #195, #196.
Activity
hyphaltip commentedon Nov 9, 2026
Curator decision 2026-11-09 (grilling on the v0.9.0 gate causes): Saccharomyces uvarum is classed
NOT_TYPABLE:silent_cassetteslike the rest of its genus. Its panel is an unexplained exception: it gives single calls in 263 of 294 runs, at flank depth 1.7, with no assembly comparison, while the other Saccharomyces panels givenoneorboth. The site hides its allele counts.Part of this issue: explain those calls. One candidate is k-mers that differ between the cassette copies, not the active locus. The ratio test (Ya:Yalpha depth) will show whether they carry real signal. Evidence:
datasets/reads/sra_survey/pilot/cause_evidence.tsvandcauses.tsv(branchv090-gate-causes).