test(sv): prove the TRExplorer rule regions and the jellyfish sizing in CI, and correct the AnnotSV row count

.github/workflows/e2e.yml

@@ -637,7 +637,7 @@ jobs:
           grep -qx archaic-ancestry-1 "$list" && echo "data_archaic=true" >> "$GITHUB_OUTPUT"
           grep -qx bash-step-08-hla "$list" && echo "data_hla=true" >> "$GITHUB_OUTPUT"
           grep -qx prs-scores-and-run-options-1-strand-sex "$list" && echo "data_pgs=true" >> "$GITHUB_OUTPUT"
-          grep -qxE 'callability-1-catalog|cn-intervals-1-catalog' "$list" && echo "data_catalogs=true" >> "$GITHUB_OUTPUT"
+          grep -qxE 'callability-1-catalog|cn-intervals-1-catalog|followups-ci-3-sv-rule-regions' "$list" && echo "data_catalogs=true" >> "$GITHUB_OUTPUT"
           echo "list=${list}" >> "$GITHUB_OUTPUT"
 
       # Before the base outputs: the base leg pulled the same images.
@@ -693,10 +693,10 @@ jobs:
       # of two IPD-IMGT/HLA releases and GENCODE's genes and exons
       # (bash-step-08-hla, which then runs with the network blocked), and the
       # four real PGS Catalog score files of prs-scores-and-run-options-1.
-      # A third keyed the same way: the pinned downloads of the two table
+      # A third keyed the same way: the pinned downloads of the three table
       # builders (GENCODE's GTF from EBI among them) that
-      # callability-1-catalog and cn-intervals-1-catalog rebuild their
-      # tables from.
+      # callability-1-catalog, cn-intervals-1-catalog and
+      # followups-ci-3-sv-rule-regions rebuild their tables from.
       - name: Data cache keys
         id: data-keys
         run: |

docs/05-annotsv.md

@@ -124,7 +124,7 @@ Some class 4 and 5 calls are common or unreliable for a reason AnnotSV does not
 - **The MHC** (`mhc`). An SV more than half inside the MHC, GRCh38 chr6:28,510,120-33,480,577. Its haplotypes differ by tens of kilobases between people, and gnomAD-SV marks every SV over half inside IGH or the MHC as low confidence (FILTER `IGH_MHC_OVERLAP`, Collins 2020, PMID 32461652). CYP21A2 lies in the MHC; short reads cannot call it here, and [step 35b](35-paralogs.md) reads it from PacBio HiFi reads.
 - **A pseudogene named by a disease gene's symbol** (`gene_alias`). AnnotSV joins its gene annotations by symbol. A few pseudogenes carry a disease gene's symbol as an HGNC alias or previous symbol (for example RHEBP1, whose HGNC alias is RHEB). When `Gene_name` lists that symbol and the SV reaches no transcript of the disease gene in AnnotSV's genes file, the table names the pseudogene instead, and the call moves to class 3 when no other gene is left. This needs the genes file, like the exon-edge check.
 
-The regions are in `assets/sv_rule_regions.tsv`, one row each (`chrom`, `start` 0-based, `end`, `rule`, `action`, `name`, `source`): 2,349 coding tandem repeats, the six IG and TR loci, the MHC and 10 pseudogenes. `scripts/ci/build-sv-rule-regions.sh` writes it from four pinned downloads, each checked against its sha256: TRExplorer v2.1's GRCh38 tandem repeat catalog (every repeat of 50 bp or more on a MANE Select coding exon of GENCODE 50), GENCODE 50's IG and TR genes, the MHC row of the GRCh38 assembly regions report, and HGNC's complete set of 2026-10-06 (each pseudogene whose alias or previous symbol is the approved symbol of a gene with a GenCC curation). The file is about 250 KB and ships with the repository, so the step downloads nothing. Step 05 mounts it into the AnnotSV image and the Nextflow `ANNOTSV` gets it as an input.
+The regions are in `assets/sv_rule_regions.tsv`, one row each (`chrom`, `start` 0-based, `end`, `rule`, `action`, `name`, `source`): 2,409 coding tandem repeats, the six IG and TR loci, the MHC and 10 pseudogenes. `scripts/ci/build-sv-rule-regions.sh` writes it from four pinned downloads, each checked against its sha256: TRExplorer v2.1's GRCh38 tandem repeat catalog (the file upstream rebuilt in place on 2026-10-30, sha256 `df9c23c4…`; every repeat of 50 bp or more on a MANE Select coding exon of GENCODE 50), GENCODE 50's IG and TR genes, the MHC row of the GRCh38 assembly regions report, and HGNC's complete set of 2026-10-06 (each pseudogene whose alias or previous symbol is the approved symbol of a gene with a GenCC curation). The file is about 250 KB and ships with the repository, so the step downloads nothing. The e2e case `followups-ci-3-sv-rule-regions` rebuilds it with `scripts/ci/build-sv-rule-regions.sh --check`, which fails when one byte differs, so a hand edit or a drifted pin cannot ship. Step 05 mounts it into the AnnotSV image and the Nextflow `ANNOTSV` gets it as an input.
 
 How many callers saw each SV comes from step 22's consensus (`sv_merged/<sample>_sv_consensus.vcf.gz`), which runs after this step, and beside `ANNOTSV` in Nextflow. So the reports (`bin/collect_summary.py`) join it when they are written: an SV matches the consensus record of the same type whose two breakpoints lie within 1,000 bp of its own, step 22's distance, and its support is that record's `SUPP` with the callers of `SUPP_VEC`. An SV in no consensus record was made by one caller and shows as "single caller". Without the consensus file the support is "not known", never 1. `annotsv_postfilter.py --consensus` adds the same `Support` and `Support_callers` columns to the table.
 

docs/53-locityper.md

@@ -32,7 +32,7 @@ Pinned in `versions.env`; [Image versions](versions.md) lists the current tags.
 
 The panel is the HPRC release 2 Minigraph-Cactus pangenome against GRCh38, as a VCF (`LOCITYPER_PANEL_URL`, `LOCITYPER_PANEL_VERSION` in `versions.env`). It holds the two haplotypes of each of its 232 samples. The full VCF is 6.7 GB; `setup.sh` reads only the loci and 250 kb either side, over https, through the VCF's `.tbi`, whose sha256 `LOCITYPER_PANEL_TBI_SHA256` pins. Nothing of it is committed. The pin covers the index, not the records: they are trusted as the HPRC's S3 bucket serves them over https, as for the other data `setup.sh` reads in place. A changed file with the same index would build without an error.
 
-Then it follows Locityper's documentation (https://locityper.vercel.app, "Preparing targets"): jellyfish counts the 25-mers of `REF_FASTA` (its hash table sized to the reference's length from the `.fai`, at least 10 million entries, so a small reference does not reserve the memory all of GRCh38 needs), `locityper target` cuts the panel's haplotypes at each locus, and `locityper augment` aligns them to each other, which the weighted distance needs. A new reference, image, panel or locus list builds again. `LOCITYPER_DATA` puts the database elsewhere and `LOCITYPER_LOCI` replaces the locus list.
+Then it follows Locityper's documentation (https://locityper.vercel.app, "Preparing targets"): jellyfish counts the 25-mers of `REF_FASTA` (its hash table sized to the reference's length from the `.fai`, at least 10 million entries, so a small reference does not reserve the memory all of GRCh38 needs: on the 235 Mb reference of the e2e case `ci-e2e-2-locityper` the count took 36 minutes with a table fixed at 3G entries and takes 36 seconds now, and the case went from 2,289 to 142 seconds), `locityper target` cuts the panel's haplotypes at each locus, and `locityper augment` aligns them to each other, which the weighted distance needs. A new reference, image, panel or locus list builds again. `LOCITYPER_DATA` puts the database elsewhere and `LOCITYPER_LOCI` replaces the locus list.
 
 ## The loci