Repository navigation
Yeast genomes with loci from two MAT systems: route Ogataea to MATog, re-run 181 phylum-fallback genomes #188
Description
Problem
In the v0.7.0 clean run, 181 Saccharomycotina genomes have loci called by two or more of the families MATsc, MTL, MATog and MATkp. All 181 were routed by phylum_fallback, so every Ascomycota family was tried; none of the lineage- or direct-routed genomes is affected (results/2026-10-28_idiomorph_crosscall_flags/crosscall_genomes.tsv, made by scripts/idiomorph_crosscall_flags.py).
- 101 genomes have only duplicate calls: the second family's locus has the same contig, start and end as a locus of the first (one locus called by two panels).
- Pichiales is the main group: 72 of 112 Ogataea genomes and 157 of 158 Pichia genomes are phylum-fallback, although the MATog family (and MATkp for Komagataella) exists.
- Worked example, Ogataea glucozyma GCA_003706155.2:
PPKO02000069.1:20-23514is called as MATsc MATalpha and as MTL alpha, bothhigh,idiomorph_gene_only, at 30-47% identity; MATog was not tried.
Proposed
- Route Ogataea genomes to MATog and give Pichia and relatives a suitable family (or a Pichiales-level route); curator ruling on the routing.
- Re-run the 181 genomes and compare calls and counts with v0.7.0.
- Until then, the output-level rule
scripts/dedupe_family_calls.pymarks the duplicate calls (185 loci in 164 genomes in v0.7.0); it changes no pipeline output.
Changes calls, so not for a labelling release. Suggested milestone v0.8.0 (not created). Related: the naming proposal docs/notes/2026-10-28_idiomorph-naming-proposal.md.
Activity
hyphaltip commentedon Oct 29, 2026
Added after review (2026-10-28)
Test plan
PPKO02000069.1:20-23514) must be called by MATog only after the routing change, not by MATsc and MTL at identical coordinates.scripts/dedupe_family_calls.pymarks 185 loci in 164 genomes.Decision needed first: the routing for Pichia and relatives (a Pichiales-level route or a new family), which depends on the Pichia reference in #189. Milestone set to v0.8.0 (it changes calls).
hyphaltip commentedon Oct 29, 2026
Routing experiment finished on 354 yeast genomes (PR #213). v1 (MTL not searched outside Serinales) removes the cross-call duplicates (181 to 35 genomes with two loci) but loses all 22 MTL-only calls. v2 (v1 plus Ogataea to MATog) moves 44 Ogataea genomes from MATsc to MATog, including 9 in the MATsc-only control. The no-call control stays empty. Needs a ruling on whether the 22 MTL-only calls are genuine; details in results/2026-10-29_yeast_routing/README.md.
hyphaltip commentedon Oct 29, 2026
Rule implemented in PR #214 (--cross-family-policy, off by default). Replay on the 354 genomes: no MTL-only call lost (5 family_confirmed, 17 family_unconfirmed), cross-called genomes with two or more loci 181 to 45, controls unchanged. 22 genomes where MTL wins on score are listed for review in PR #213's README.
detect: cross-call tie decided by best core identity (near tie keeps …
Merge pull request #213 from stajichlab/yeast-routing
Merge pull request #214 from stajichlab/family-unconfirmed
hyphaltip commentedon Oct 30, 2026
Status for v0.8.0 (2026-10-30).
Landed (merged): #213 (the routing experiment), #214 (
--cross-family-policy {off,on}: cross-calls resolved, out-of-scope MTL calls flaggedfamily_unconfirmed/family_confirmed, MATog cassette flag), #215 (browser examples).Open PRs that finish the issue:
family_ambiguouson every unscoped cross-call), plus the flank rule (an out-of-scope MTL call with fewer than two of PAP1/OBP1/PIK1 cannot win against a family whose own flank gene, SLA2 or DIC1, is in the call). On the 156 cross-called pairs of the v0.7.0 replay, about 40 of the 51 MTL-kept pairs flip to MATsc under the rule (estimate from the pair table, not a replay).registry_version, frozen codes (PicKud,SchPom,SchCom, and nowSacSch,PetFab,BarCal),label_statuswith thefamily_unconfirmedvalue.Still to do: the v0.8.0 replay with
--preset v0.8(the 99 Ogataea genomes, the 56 P. kudriavzevii genomes and the other phylum-fallback yeasts), including the lost call in GCA_019232045.1 (Ogataea scope) and the two uncalled P. kudriavzevii genomes (GCA_001879685.1, GCA_029290715.1).hyphaltip commentedon Oct 31, 2026
Status check on main (6fd981b) against the v0.7.0 baseline, 181 cross-call genomes:
Closing as done for v0.8.0.
🤖 Generated with Claude Code