hyphaltip · Oct 4, 2026 · Oct 4, 2026 · Oct 4, 2026 · Oct 4, 2026 · Oct 4, 2026
diff --git a/analysis/2026-10-03_mucoromycotina-mat-campaign.md b/analysis/2026-10-03_mucoromycotina-mat-campaign.md
@@ -0,0 +1,59 @@
+# Mucoromycotina MAT campaign: calls, locus size, synteny, sexP/sexM trees (2026-10-03)
+
+Status: running (HMG-box tree and RAxML-NG check not in yet); figures are drafts.
+
+## Question
+Across BFD, LCG and Jena Mucoromycotina genomes: which genomes carry a sexP
+(Plus) or sexM (Minus) MAT locus, how large is the locus per genus and
+idiomorph, how is gene order kept across genera, and do sexP and sexM form
+separate clades in a rooted gene tree?
+
+## Ruling
+J. Stajich, 2026-10-03: run Mucoromycotina first; locus size = inner ends of
+the flanking genes; root the gene tree with a non-MAT HMG-box outgroup; a
+synteny figure of about 20-30 pruned loci.
+
+## Data
+- 974 genomes: BFD Mucorales + Umbelopsidales 289 (288 after suppress), LCG 621,
+  Jena 64. LCG and Jena are held-out sets (no training, curation or classifier
+  use). Code: v0.6.0 (`run-7c7ed99`).
+- Taxon overrides (`db/taxon_overrides.tsv`): confirmed rows are counted under
+  the likely genus; `hold out` and `unconfirmed` rows are dropped from locus
+  size and the tree set.
+
+## Method
+See `results/2026-10-03_mucoromycotina_mat/NOTE.md` for every step and file.
+
+## Results
+- Detection: 865 of 973 genomes called; 911 loci (Plus 482, Minus 425,
+  undetermined 4). BFD 256/288, LCG 548/621, Jena 61/64. 45 genomes have more
+  than one locus; 34 are flagged `two_idiomorphs`.
+- Locus size (579 loci, 23 genera, polished flanks on both sides). Largest:
+  Umbelopsis (median Plus 13,063 / Minus 12,256 bp) and Absidia (9,611 / 9,249).
+  Intermediate: Phycomyces (6,531 / 4,166), Backusella (6,113 / 6,057), Pilaira
+  (4,726 / 2,968). Most Mucoraceae and Rhizopus: about 1.4-2.6 kb. The flank
+  pair is not the same in every lineage, so sizes compare one flank interval per
+  lineage, not one fixed gene set.
+- Synteny: 26 loci (Plus and Minus for 13 genera). Figure:
+  `synteny/mucoromycotina_MAT_synteny.png`; interactive:
+  `mucoromycotina_MAT_synteny.clinker.html.zst`.
+- Full-length gene tree (526 sequences x 526 sites, JTT+F+R8, UFBoot 1000):
+  - sexP clade: 163 tips, 0 sexM tips inside, UFBoot 40.
+  - sexM clade: 173 tips, 0 sexP tips inside, UFBoot 93.
+  - 3 outgroup HMG genes from Mucorales genomes fall inside these clades (2 in
+    sexP, 1 in sexM); they may be MAT-gene paralogs away from the locus.
+  - The outgroup is not monophyletic; 132 outgroup tips fall between sexP and
+    sexM after rooting. This tree does not support or reject sexP + sexM as
+    sister groups.
+
+## Decision
+None yet. Figures are drafts for the curator.
+
+## Open
+- HMG-box tree (job 29389343) and RAxML-NG check.
+- Compare locus size within one flank pair, or show all with the pair marked?
+- May held-out LCG/Jena genomes appear in publication figures? (They are used
+  here for description only, not for any build or score.)
+- LCG called 548 genomes on v0.6.0 versus 536 on 52b3ff9; not yet compared
+  call by call.
+- Identity of the 3 outgroup tips inside the sexP/sexM clades.
diff --git a/analysis/2026-10-04_basidiomycota-ascomycota-v060-campaign.md b/analysis/2026-10-04_basidiomycota-ascomycota-v060-campaign.md
@@ -0,0 +1,131 @@
+# Basidiomycota and Ascomycota campaign on v0.6.0 (2026-10-03/04)
+
+Status: decided (runs complete); R. toruloides HD locus open.
+
+## Question
+Can MATPredict v0.6.0 run over every BFD Basidiomycota and Ascomycota genome?
+What call rates, run times and failures does it give, and what changed against
+the 2026-09-26 Basidiomycota run?
+
+## Ruling
+J. Stajich, 2026-10-03: run a regression check, then a pilot per phylum, then
+all Basidiomycota, then all Ascomycota in waves, on the `exfab` partition.
+2026-10-04: fix the code-26 failure on a branch with a PR and v0.6.1; commit
+these results with the Mucoromycotina campaign.
+
+## Data
+- Code: frozen worktree `run-7c7ed99`. Its `src/` and `db/` are identical to
+  tag `v0.6.0`.
+- Genomes: BFD `samples.csv`, every row of the phylum with a taxid and a genome
+  file. Routing comes from the taxid (lineage routing).
+- Partition `exfab`. Accounts: `stajichlab` (384 CPUs, 1 TB shared with all
+  other jobs of the user) and `exfab` (32 CPUs, 256 GB).
+
+## Method
+1. Regression: the fixed panel (166 genomes) plus Zygo 23 on v0.6.0, diffed
+   against the last checked candidate (2026-10-01 scope-only run).
+2. Pilots: 163 Basidiomycota genomes, stratified by order (57 orders, >= 2
+   each); 246 Ascomycota genomes, stratified by class (30 classes, 92 orders).
+   `random.seed(20261003)`.
+3. Full runs: Basidiomycota as 3 jobs of 64 CPUs (genomes < 500 Mb) and 1 job
+   of 32 CPUs (58 genomes >= 500 Mb, 4 h per-genome limit). Ascomycota as 9
+   waves of about 2,157 genomes, 64 CPUs each, 1 h per-genome limit.
+
+## Results
+
+### Regression
+0 changed loci in all five panels. Zygo 23: 23/23 on scaffold and contig
+inputs. `results/2026-10-03_regression_v060/diff/summary.md`.
+
+### Basidiomycota (3,275 listed)
+- 3,270 reports; 5 on the suppress list; 0 failures; 0 timeouts.
+- Wall time: 3 jobs of about 1 h, 1 job of 32 min. Median per genome 101 s
+  (< 50 Mb) to 716 s (> 500 Mb).
+
+| Subphylum | Genomes | Called v0.6.0 | Without PR-only genomes | Called 2026-09-26 |
+|---|---|---|---|---|
+| Agaricomycotina | 2,417 | 2,231 (92.3%) | 2,021 (83.6%) | 2,007 (83.0%) |
+| Pucciniomycotina | 486 | 391 (80.5%) | 371 (76.3%) | 64 (13.2%) |
+| Ustilaginomycotina | 315 | 308 (97.8%) | 301 (95.6%) | 301 (95.6%) |
+| Wallemiomycotina | 51 | 51 (100%) | 51 (100%) | 0 |
+
+- Gains come from the records curated since 2026-09-26: `redHD`/`redPR`
+  (Sporidiobolales, about 246 genomes), `rustHD` (107), `wallMAT` (51).
+- `Basidiomycota:PR` is called in 1,111 more genomes. 1,360 PR calls carry
+  `verification: unverified` (admitted only through the strict-CAAX scan). In
+  phylum-fallback orders most calls are PR only: Sebacinales 12/12 PR-only,
+  Cantharellales 73/90, Trichosporonales 50/60, Auriculariales 18/19. Do not
+  count these as MAT loci without the CAAX caveat.
+- Apparent losses are mostly merges of one interval into one call: HD into
+  Aalpha 96, HD into redHD 15, Schizophyllum Balpha + Bbeta into one B call 40.
+- 26 Sporidiobolales (R. toruloides) genomes lost a generic-HD call (HD1/HD2,
+  often with MIP1, on the PR contig, about 200 kb from PR). v0.6.0 calls
+  `redHD` on a different contig, without MIP1. Which one is the MAT-linked HD
+  pair is open (see open questions).
+- 2 genomes lost every call: GCA_030280995.1 (Pucciniales, bLocus) and
+  GCA_054785265.1 (Kriegeriales, HD).
+- Tables: `results/2026-10-03_basidiomycota_v060/genomes.tsv`, `loci.tsv`,
+  `by_order.tsv`, `by_order_nonPR.tsv`, `misses_by_order.tsv`, `size_bins.tsv`.
+
+### Ascomycota (19,415 listed)
+- 19,380 reports; 15 on the suppress list; 20 failures. All 20 are Alaninales
+  (genetic code 26): exonerate has no table 26 and exits 1, which ended the
+  genome. Fixed in v0.6.1 (PR #13). Re-run of the 20 on v0.6.1
+  (`results/2026-10-04_alaninales_v061/`): 20/20 reports, 0 failures, 30-35 s
+  each, miniprot-only models. 4 called (P. tannophilus GCA_001661245.1: MATsc
+  MATa, MTL alpha, MAT MAT1-1; 3 others one MATsc call each), 16 uncalled. All
+  route `phylum_fallback`: no Alaninales record exists (reference gap).
+  Curator (2026-10-04): miniprot models genes less well than exonerate, so pass
+  exonerate the NCBI table as a 64-letter string (accepted by stock exonerate
+  2.4.0) instead of skipping it. Re-run on that code (`run-b04a77b`,
+  `results/2026-10-04_alaninales_v061_exostring/`): 20/20 reports, **15/20
+  called** (miniprot-only 4/20); median 457 s per genome (34 s). Regression vs
+  v0.6.0: 0 changed loci (`results/2026-10-04_regression_v061s/`). Family labels
+  under phylum fallback are not reliable (MATsc and MTL in one genome).
+- 17,314 of 19,380 genomes have >= 1 call (89.3%). 848 CPU-h; median 162 s per
+  genome; maximum 3,095 s.
+
+| Class | Genomes | Called | Routing |
+|---|---|---|---|
+| Sordariomycetes | 6,017 | 93.0% | lineage |
+| Eurotiomycetes | 3,221 | 96.5% | lineage |
+| Pichiomycetes | 2,967 | 89.9% | lineage 2,367 / fallback 575 |
+| Dothideomycetes | 2,731 | 89.4% | lineage |
+| Saccharomycetes | 2,648 | 85.1% | lineage 2,281 / fallback 364 |
+| Lecanoromycetes | 479 | 86.2% | lineage |
+| Leotiomycetes | 435 | 92.2% | lineage |
+| Dipodascomycetes | 398 | 28.7% | mostly fallback |
+| Pezizomycetes | 172 | 79.7% | lineage |
+| Orbiliomycetes | 67 | 9.0% | fallback |
+| Lipomycetes | 59 | 94.9% | fallback |
+| Taphrinomycetes | 22 | 22.7% | lineage |
+
+- 21,011 loci: Ascomycota:MAT 12,628; MATsc 4,954 (includes HML/HMR silent
+  copies); MTL 2,799; MATtub 200; PM 138; MATyl 123; mat1 100; mat2 69.
+- Confidence: high 13,740, medium 7,112, low 159. Strict pass 20,695, relaxed
+  316. `homothallic_candidate` loci: 357.
+- Ascomycota:MAT genomes: MAT1-1 only 6,269; MAT1-2 only 5,692; both 204.
+- Tables: `results/2026-10-03_ascomycota_v060/genomes.tsv`, `loci.tsv`,
+  `by_class.tsv`.
+
+## Decision
+Results kept as the v0.6.0 baseline for both phyla. Code-26 fix shipped as
+v0.6.1 (PR #13).
+
+## Open
+- R. toruloides HD: old generic-HD locus (PR contig) versus v0.6.0 `redHD`
+  locus (other contig). Examples: GCA_000222205.2 (AEVR02000017.1:693,823-734,526
+  old; AEVR02000008.1:1,252,021-1,268,991 new), GCA_001542265.1,
+  GCA_000258745.1. Curator ruling pending.
+- Classes with low call rates under phylum fallback (Dipodascomycetes,
+  Orbiliomycetes, Trigonopsidomycetes) are reference gaps, not measured
+  absence.
+- The genome with 11 Ascomycota loci has not been examined.
+
+## Files
+- `results/2026-10-03_regression_v060/`
+- `results/2026-10-03_basidiomycota_pilot/`, `results/2026-10-03_ascomycota_pilot/`
+- `results/2026-10-03_basidiomycota_v060/`, `results/2026-10-03_ascomycota_v060/`
+  (`reports_all.tar.zst` = every `detection_report.yaml` and `wall_seconds`;
+  per-genome GFF3 and FASTA are not committed)
+- `results/2026-10-04_alaninales_v061/`
diff --git a/analysis/INDEX.md b/analysis/INDEX.md
@@ -15,15 +15,17 @@ in yet; superseded = replaced by a later study. `results/` = main checkout.
 | 09-29 | [Classifier builds: aligner, determinism, gate, P1](2026-09-29_classifier-builds.md) | decided; shipped 9c39e2c | no aligner more accurate; mafft --auto = L-INS-i; ClipKIT 16 wrong; builds byte-identical; P1 withholds 26, reveals 5 |
 | 09-29 | [sexM-like paralogs, strain labels, Absidia](2026-09-29_paralogs-and-strain-labels.md) | decided | labels 10/7/4 (n=21); P1 in both mating types; Absidia flanks off-scaffold |
 | 09-29 | [Two idiomorphs and homothallism](2026-09-29_two-idiomorphs-and-homothallism.md) | decided (report-only); causes open | 36 unlinked; literature signal 13/16 vs 5/9; Syzygites both idiomorphs |
+| 10-03 | [Mucoromycotina MAT campaign](2026-10-03_mucoromycotina-mat-campaign.md) | running (HMG tree) | 865/973 genomes, 911 loci; locus size 579 loci/23 genera; full-length tree sexP 0 sexM inside (UFBoot 40), sexM 0 sexP inside (UFBoot 93) |
 
 ## Basidiomycota
 | Date | Report | Status | Key numbers |
-| 10-01 | [Branch merges and scope-only families](2026-10-01_basidio-merges-and-scope-only.md) | decided | net 14 gained in-lineage, 2 CAAX-only PR lost, 0 label changes |
 |---|---|---|---|
+| 10-01 | [Branch merges and scope-only families](2026-10-01_basidio-merges-and-scope-only.md) | decided | net 14 gained in-lineage, 2 CAAX-only PR lost, 0 label changes |
 | 09-26 | [Anchors, full run, cap-off](2026-09-26_basidiomycota.md) | decided | 3,269 genomes; Agaricomycotina 83.0% |
 | 09-26 | [Cryptococcus SXI slot](2026-09-26_cryptococcus-sxi.md) | decided | 243 genomes, 0 lost, 53 gained |
 | 09-26 | [Pucciniomycotina, Wallemia, Rhodotorula](2026-09-26_pucciniomycotina-wallemia-rhodotorula.md) | decided | Wallemia 51/51; Rhodotorula P/R 61/62 held-out |
 | 09-27 | [Receptor (B/PR) loci](2026-09-27_receptor-loci.md) | decided (review later) | CAAX: 6/9 vs 0/25; Agaricales PR 14->132 |
+| 10-04 | [Basidiomycota + Ascomycota v0.6.0 campaign](2026-10-04_basidiomycota-ascomycota-v060-campaign.md) | decided; R. toruloides HD open | Basidio 2,982/3,270 (2,745 without PR-only); Asco 17,314/19,380; 20 code-26 failures -> v0.6.1 |
 
 ## Ascomycota (incl. Serinales, Dothideomycetes)
 | Date | Report | Status | Key numbers |
@@ -33,6 +35,7 @@ in yet; superseded = replaced by a later study. `results/` = main checkout.
 | 09-26 | [Serinales, C. albicans, C. auris](2026-09-26_serinales-candida.md) | decided | C. albicans 8/11 collapsed |
 | 09-26 | [Dothideomycetes curation](2026-09-26_dothideomycetes.md) | decided | 90/99; C. kikuchii homothallism candidate |
 | 09-27 | [Confidence rules](2026-09-27_confidence-rules.md) | decided | 86 risers; no fallback floor |
+| 10-04 | [Basidiomycota + Ascomycota v0.6.0 campaign](2026-10-04_basidiomycota-ascomycota-v060-campaign.md) | decided | 19,415 genomes; 89.3% called; Dipodascomycetes 28.7%, Orbiliomycetes 9.0% (fallback) |
 
 ## Validation and held-out sets
 | Date | Report | Status | Key numbers |

diff --git a/analysis/open-questions.md b/analysis/open-questions.md
@@ -74,3 +74,13 @@ not match the genome's name:
 | Item | Next step | Where |
 |---|---|---|
 | S. pombe P and Yarrowia A/B idiomorph class (unassigned) | protein-domain evidence (Pc vs alpha box; MATA/MATB proteins) | `db/Ascomycota/order.yml` (B9) |
+
+## Opened 2026-10-04 (campaign runs)
+
+| Item | Next step | Where |
+|---|---|---|
+| R. toruloides HD: 2026-09-26 generic-HD locus (PR contig, ~200 kb from PR, with MIP1) vs v0.6.0 `redHD` locus (other contig, no MIP1); 26 genomes | curator ruling on which is the MAT-linked HD pair | `2026-10-04_basidiomycota-ascomycota-v060-campaign.md` |
+| Basidiomycota PR calls: 1,360 unverified (strict-CAAX only); most phylum-fallback calls are PR only | keep the caveat in any summary; review with the receptor-loci study | same |
+| Mucoromycotina locus size: flank pair differs by lineage | compare within one pair, or plot all with the pair marked | `2026-10-03_mucoromycotina-mat-campaign.md` |
+| Held-out LCG/Jena genomes in publication figures | curator ruling | same |
+| 3 outgroup HMG genes inside the sexP/sexM clades (Dicele1 h5, GCA_016758965.1 h6, Mycafr1 h1) | check locus and identity | same |
diff --git a/results/2026-10-03_ascomycota_pilot/README.md b/results/2026-10-03_ascomycota_pilot/README.md
@@ -0,0 +1,7 @@
+# Ascomycota pilot on v0.6.0 (2026-10-03)
+
+246 BFD Ascomycota genomes, stratified by CLASS (30 classes, 92 orders): about 200 x class
+share, at least 3 per class, round-robin over orders inside a class (random.seed 20261003).
+Lists: pilot_list.tsv, pilot_meta.tsv. Code: frozen worktree run-7c7ed99 (= v0.6.0).
+Partition exfab, 32-way, GENOME_TIMEOUT 3600 (no genome > 500 Mb in the pick).
+Purpose: per-genome runtime and call rate per class before the 19,415-genome run.
diff --git a/results/2026-10-03_ascomycota_pilot/jobs.tsv b/results/2026-10-03_ascomycota_pilot/jobs.tsv
@@ -0,0 +1 @@
+29387476	pilot	246
diff --git a/results/2026-10-03_ascomycota_pilot/pilot_by_stratum.tsv b/results/2026-10-03_ascomycota_pilot/pilot_by_stratum.tsv
@@ -0,0 +1,31 @@
+stratum	genomes	reports	called	routing	wall_median_s	wall_p90_s	wall_max_s
+Sordariomycetes	62	62	52	lineage:62	159	213	277
+Eurotiomycetes	33	33	29	lineage:33	156	211	466
+Pichiomycetes	31	21	13	lineage:11; phylum_fallback:10	52	324	425
+Dothideomycetes	28	27	21	lineage:27	112	184	229
+Saccharomycetes	27	27	18	phylum_fallback:20; lineage:7	219	345	494
+Lecanoromycetes	5	5	4	lineage:5	139	164	164
+Dipodascomycetes	4	4	2	phylum_fallback:2; direct:2	139	358	358
+Leotiomycetes	4	4	4	lineage:4	197	219	219
+Alloascoideomycetes	3	3	3	phylum_fallback:3	181	311	311
+Arthoniomycetes	3	3	2	lineage:3	103	170	170
+Candelariomycetes	3	3	3	lineage:3	150	238	238
+Geoglossomycetes	3	3	2	lineage:3	171	183	183
+Lichinomycetes	3	3	3	lineage:3	170	186	186
+Lipomycetes	3	3	3	phylum_fallback:3	549	581	581
+NA	3	3	3	phylum_fallback:2; lineage:1	317	624	624
+Orbiliomycetes	3	3	0	phylum_fallback:3	175	203	203
+Pezizomycetes	3	3	2	lineage:3	33	50	50
+Pneumocystomycetes	3	3	3	lineage:3	5	5	5
+Schizosaccharomycetes	3	3	3	direct:3	11	34	34
+Sporopachydermiomycetes	3	3	3	phylum_fallback:3	278	303	303
+Taphrinomycetes	3	3	1	lineage:3	5	5	5
+Trigonopsidomycetes	3	3	0	phylum_fallback:3	250	322	322
+Coniocybomycetes	2	2	2	lineage:2	124	132	132
+Saitoellomycetes	2	2	0	lineage:2	7	8	8
+Laboulbeniomycetes	1	1	0	lineage:1	52	52	52
+Neolectomycetes	1	1	0	lineage:1	5	5	5
+Novakomycetes	1	1	0	lineage:1	3	3	3
+Sareomycetes	1	1	1	lineage:1	171	171	171
+Xylobotryomycetes	1	1	1	lineage:1	83	83	83
+Xylonomycetes	1	1	1	lineage:1	123	123	123