Reviewed GOA, UniProt Q6QA32, PMID:16216911, and PANTHER family cache. The direct core annotation is endoplasmic-reticulum Hsd17b3 testosterone dehydrogenase activity in androgen/testosterone biosynthesis.
The whole review was templated: every annotation carried the same UniProt FUNCTION line as supporting_text, including the three localization terms. All eleven rows were rewritten against the primary paper and the UniProt record, and the three PENDING rows were resolved.
ZFIN remapping. Current QuickGO attaches ZFIN's three experimental rows (GO:0005783 IDA, GO:0006702 IDA, GO:0047045 IDA, all from PMID:16216911) to a newer TrEMBL accession of the same zebrafish gene, so they are absent from the refreshed hsd17b3-goa.tsv obtained under Q6QA32. They are legitimate annotations of this gene product and were retained and reviewed on their merits; this is noted in the summary of each of those rows.
Evidence used. PMID:16216911 is abstract-only in the cache, but the abstract states each of the facts relied on:
- catalysis: PMID:16216911
- substrate scope: PMID:16216911 and PMID:16216911 — this is the route toward 11-ketotestosterone, the principal teleost androgen
- localization: PMID:16216911
- expression: PMID:16216911
PMID:27927697 (Tsachaki et al., J Endocrinol 2017) was newly cached and added to references:. UniProt cites it (ECO:0000269) for the EC 1.1.1.64 reaction and for 11-ketoandrostenedione reduction; its abstract does not name hsd17b3, so it is used only as corroboration, never as sole support, and reference_review records that.
Action changes.
- GO:0016020 membrane (IEA, SL-0162): KEEP_AS_NON_CORE -> MODIFY to GO:0005789 endoplasmic reticulum membrane. Experimental ER localization plus the UniProt-predicted helical TRANSMEM at residues 6-26 supports the specific child.
- GO:0005737 cytoplasm (IEA GO_REF:0000117) and a duplicate GO:0005783 (IEA GO_REF:0000044) were removed from the yaml: validation reported both as "not in GOA", i.e. they are no longer in the refreshed record. No other row lost.
- GO:0008610 lipid biosynthetic process (IEA, ARBA00027236, PENDING): MODIFY to GO:0006702 androgen biosynthetic process. True but a very distant ancestor; the specific descendants are already annotated by IDA.
- GO:0033764 steroid dehydrogenase activity, CH-OH donors, NAD/NADP acceptor (IEA, ARBA00084320, PENDING): MODIFY to GO:0047045. The exact reaction is experimentally established, so the grouping parent adds nothing.
- GO:0005783 endoplasmic reticulum (IEA GO_REF:0000120, PENDING): ACCEPT, concordant with the IDA.
- The two IBA rows (GO:0005783 is_active_in, GO:0047045 enables) were kept as ACCEPT with reasoning about node placement rather than donor count. ZFIN:ZDB-GENE-040426-1339 appears in both WITH/FROM lists because this gene's own IDA seeded the IBD — expected, not circular.
- The three IDA rows remain ACCEPT; GO:0006702 passes the participation test because the enzyme itself catalyses a step of androgen synthesis rather than merely being required for it.
Definitions verified via QuickGO REST for GO:0033764, GO:0008610, GO:0047045, GO:0061370, GO:0006702, GO:0005789 and GO:0005783; none obsolete.
Caveat recorded, not acted on. The deep-research file reports that hsd17b3 transcripts are low or undetected in juvenile gonads at 19 and 30 dpf, so another Hsd17b may supply the 17beta-reduction at those stages. This tempers the in vivo weight of the testosterone-biosynthesis annotation but does not contradict the demonstrated enzymology, so GO:0061370 stays ACCEPT with the caveat in reason, and the question is raised in suggested_questions.
description was rewritten as standalone biology (SDR fold, ER anchor, NADPH-dependent 17beta-reduction, substrate scope including 11-oxygenated androgens, non-testis-restricted expression). Validation: zero errors.