The 21 seeded assertions and their complete source objects are preserved. All three official products were reconstructed independently from the VSP changes and official FASTA: 134, 98 and 124 residues. The historical 97-residue erythrocyte protein excludes the initiator methionine; the current translated erythrocyte product has 98 residues. Membrane-anchor deletion, internal deletion and arbitrary experimental truncation are distinct. Retention of both axial histidines does not prove equal function of every product.
The review follows the project-specific protein-binding policy in CLINGEN_MENDELIAN: source-supported generic interactions remain non-core unless a verified partner class justifies refinement. The five Haenig pairs have fifteen current literal records, with all thirty backbone ROGIDs independently matching official sequences. KIF1B isoform 2 is tested as its 1192–1770 tail region, supporting kinesin binding without assigning motor activity or native transport. PMP22 is the exact 217-residue unreviewed product, progranulin is the 593-residue precursor, and RNF11 is not automatically treated as an active E3 enzyme. The four P450 interactions are explicitly positive in the 15680923 abstract, despite zero current exact IntAct/MINT joins; binding does not establish activation of all four enzymes.
The actual 1,037-node PAINT tree, target MSA and 21-node ancestral path support both literal IBD placements. PTN000451786 carries the ER-membrane inference with rat Cyb5a evidence. PTN000451724 carries the family-wide heme-binding inference with human CYB5B and zebrafish Cyb5a/Cyb5b evidence. This valid shared chemistry does not transfer a donor's mitochondrial localization. The target's own experimental evidence at other family-root terms is legitimate, not circular. Six literal source aliases were resolved against 106 current donor records. InterPro IPR018506 matches the actual human 40–47 heme-ligand motif.
Two mitochondrial-outer-membrane assertions remain uncertain. Both Reactome paths really identify P00167 residues 2–134, but the cited 8513896 review distinguishes the separate ER and mitochondrial cytochrome-b5 genes. The 7668255 accessible abstract assays reductase deficiency and ascorbate regeneration, without resolving the CYB5A location. Failed full-original access is recorded; no abstract-only wrong-gene rejection was made. The SGTA pathway instead explicitly represents full-length P00167-1 temporarily in cytosol before insertion, not the soluble erythrocyte splice form.
The 19946888 NK-cell membrane target table remains unavailable after two 403 supplementary-table routes; the HDA assertion is retained as UNDECIDED, without disputing independently supported membrane localization. The 22375059 cache contains only Abstract/Introduction despite its full-text flag. Separately accessed indexed Results/Methods explicitly describe cytb5-ops28 ER insertion, Figure 2D/Table 1 and transport assays; the exact donor-construct species/sequence is not supplied by those read passages. This supports the ER-membrane refinement while retaining the curated target identity and experimental limits. No machine cache was edited to make its access flag match the separately recovered source.
The core is a heme-dependent electron relay and partner-enzyme cofactor. Human 22170710 assays directly establish CYP17A1 lyase enhancement, while H44L heme loss is a structural prediction. The 2018 direct electron-transfer experiment 29888793 uses human CYP17A1, rat reductase and rabbit b5. It is mechanistic ortholog support, not human CYB5A-specific IDA. The 2016 Mn-b5 experiment and 2021 resonance-Raman work show that redox and substrate-dependent conformational contributions must both be retained. An exclusively allosteric explanation is not justified, and enzyme binding alone is not enzyme activation. No new process annotation was proposed from necessity, disease phenotype or absence in a pathway. No GO-CAM target activity was found in the protected index. The single core represents this integrated carrier role; it does not require a new source assertion or a duplicate enzyme-activation core.
Research used genuine default Falcon with the configured perplexity-lite fallback. Authentication/quota failures were terminal and no provider report was produced. Genuine normal gene/publication fetches succeeded; protected caches and contexts remained exact. Two generated PANTHER family context files stay TMP-only because pending PR variants own those paths; neither pending variant was adopted.
Actual reading boundaries: selected Haenig author-PDF Results and Y2H Methods; indexed 22375059 Results/transport Methods; complete available source abstracts; selected original 22170710, 27297105, 29888793 and 33656879 passages recovered by the independent core consultant and then personally reread. No modern supplementary protocols or figure pixels were inspected. HPA evidence is assay metadata only. Full raw responses, construct hashes, donor records and failed-access receipts remain in the immutable audit packet; the public source index provides compact source locators and conclusions.
One NEW GO:0009055 assertion now represents the conserved performed electron-transfer activity. This supersedes the earlier decision to describe it only in the core. It uses ISS with PMID:29888793 and rabbit CYB5A P00169, not human IDA. The original Methods identify human CYP17A1, rat reductase and full-anchor rabbit b5; stopped-flow Results identify coupled b5 oxidation and CYP17A1 oxy-complex reduction. The official human/rabbit 134-residue proteins share 121 positions and 75/77 positions in the heme-binding domain, with both axial histidines retained. This sequence support complements the measured donor activity; it does not replace an assay or authenticate an unpublished construct sequence. Human H44L coexpression assays establish impaired lyase enhancement, while heme loss was predicted. The Mn-b5 experiment separates binding from redox-dependent enhancement. Substrate-dependent conformational effects remain part of the biology.
The original 21 source objects, including absent isoform fields, are preserved. Two curated functional classes map the anchored P00167-1 and soluble P00167-2 products without retrofitting those source fields. The structurally distinct P00167-3 remains documented but is not assumed functionally equivalent. Human erythrocyte sequencing explicitly establishes cytosol, supporting refinement of the broad cytoplasm mapping. ER-membrane topology/insertion supports refinement of the broad ER transfer while preserving the mouse donor and its actual access limit.
The cytosol actions deliberately distinguish mature protein from biogenesis: the HPA assertion remains ACCEPT with independent mature soluble-protein corroboration; the SGTA source remains KEEP_AS_NON_CORE because its actual entity is the full-length tail-anchored product before membrane insertion. A structured consistency exception makes this distinction explicit. Cytosol remains in the core for the mature soluble class, not because an SGTA-bound precursor was shown to carry out redox chemistry.
The two mitochondrial rows remain UNDECIDED. Reactome 198824 attributes myristoylation to CYB5R3, whereas 198845 explicitly describes CYB5A at the outer membrane. Both actual target paths identify P00167. The cited review distinguishes ER and mitochondrial b5 genes, and the patient paper measures reductase deficiency. Neither inaccessible original is treated as a proven wrong-gene source. The modeled alcohol-dehydrogenase label is not transferred; its written semidehydroascorbate reaction is recognized as electron transfer.
The new mechanistic findings reuse the existing 2018 quote once and add a short 2016 anchor. No original quotation is silently dropped or multiplied. Screen reasons now retain partner-specific class constraints without repeating the full shared Methods. Kinesin identity supports the KIF1B refinement; the other four inspected partners do not establish a narrower class at this scope. An androgen-biosynthesis question acknowledges that electron donation performs part of the chemistry; no new process assertion is made without its separate convention and comparator assessment.