Correct the GO taxon count to 134,390, and trace every file to its upstream URL

README.md

@@ -29,6 +29,12 @@ WS298 is WormBase's **final production release**. Ongoing curation moved to the
 rather than one that will be superseded — which is the reason to pin it rather than track a moving
 `current-production-release/` alias.
 
+**Upstream:** [wormbase.org](https://wormbase.org/) · [downloads.wormbase.org](https://downloads.wormbase.org/)
+· [ONTOLOGY/ directory for this release](https://downloads.wormbase.org/releases/current-production-release/ONTOLOGY/)
+· [WormBase 2024 paper](https://doi.org/10.1093/genetics/iyae050)
+
+Every file here is traceable to a single upstream URL; see [Provenance](#provenance).
+
 ## Layout
 
 ```text
@@ -157,10 +163,15 @@ headers rather than a column row. Download and parse them directly.
 phenotype. "The genes for this phenotype" is the wrong question unless you read column 4 — a
 `NOT` qualifier — or filter on it.
 
-**Most GO records are not *C. elegans*.** Only 134,137 of `gene_association`'s 312,203 rows carry
-`taxon:6239` as their primary taxon; the rest are nine other nematodes. (134,390 rows mention
-`taxon:6239` in *either* taxon field, the extra 253 being interacting-taxon entries — a different
-question with a different answer.)
+**Most GO records are not *C. elegans*.** Only 134,390 of `gene_association`'s 312,203 rows carry
+`taxon:6239`; the rest are nine other nematodes — *C. briggsae*, *C. remanei*, *P. pacificus*,
+*B. malayi* and five more.
+
+Watch the arithmetic here. A `grep` for the bare column value returns 134,137, but 253 further worm
+rows name a second, interacting taxon and so read `taxon:6239|taxon:287` — always a pathogen
+(*P. aeruginosa*, *E. coli*, *B. thuringiensis* …), never another host. Those are *C. elegans* rows.
+No row carries `taxon:6239` in the interacting position alone, so 134,390 is the count of rows about
+the worm and 134,137 is only the count of rows about the worm *and nothing else*.
 
 **Anatomy containment is `part_of`, not `is_a`.** Walking `is_a` alone answers nothing about what
 lies inside the pharynx. Widen the relation set when traversing the anatomy ontology; in the
@@ -168,12 +179,30 @@ development ontology, note that `preceded_by` is a temporal ordering rather than
 
 ## Provenance
 
-Downloaded from `downloads.wormbase.org` on 2026-07-26. The ontology and association files came
-through the `releases/current-production-release/ONTOLOGY/` alias, which is the only path WormBase
-serves — the concrete `releases/WS298/` directory the filenames imply answers 403.
+Downloaded from [downloads.wormbase.org](https://downloads.wormbase.org/) on 2026-07-26.
+
+The twelve files in `WS298/ontology/` each came from exactly one URL, formed by appending the
+filename to the release's ONTOLOGY directory:
+
+```text
+https://downloads.wormbase.org/releases/current-production-release/ONTOLOGY/<filename>
+```
 
-Every file was verified as intact gzip and parsed end to end before upload; the record counts in
-the tables above are measured from these exact bytes, not copied from upstream documentation.
+So `anatomy_ontology.WS298.obo.gz` is
+[…/ONTOLOGY/anatomy_ontology.WS298.obo.gz](https://downloads.wormbase.org/releases/current-production-release/ONTOLOGY/anatomy_ontology.WS298.obo.gz),
+and so on for the other eleven. Note the path goes through the `current-production-release/` alias:
+that is the only path WormBase serves, and the concrete `releases/WS298/` directory the filenames
+imply answers 403. Because WS298 is the final release, the alias and the pin name the same bytes —
+but if WormBase ever advances the alias, these files remain WS298 and upstream would not.
+
+The two files in `WS298/acedb/` come from the same release's AceDB per-class dump tree. That
+directory is reachable from a browser but Cloudflare-gated to non-interactive clients, so it cannot
+be re-fetched by script — which is part of why they are mirrored here at all.
+
+Every file was verified as intact gzip and parsed end to end before upload, and re-verified
+SHA-256-identical to the source after a round trip through this repo. The counts in the tables
+above are measured from these exact bytes, not copied from upstream documentation — where the two
+disagreed, the bytes won.
 
 ## Licensing