The historical review contains 41 source assertions. All remain unchanged outside their reviewer-supplied assessments. Five additional assertion triples were genuinely retrieved from current GOA and are reviewed separately, giving 46 contexts. These are source additions rather than NEW proposals. The original GOA, UniProt and Falcon files are preserved; the contemporary normal GOA export is provided as 2026-10-17 QuickGO snapshot. Earlier curation remains available in repository history. No previous notes file existed at the input boundary.
The biological synthesis now has two coupled activities: Hsp70 ATPase activation and ubiquitin-dependent client recognition/triage. The former broad folding core repeated those mechanisms and overstated the assigned reporter experiments. PMID:21231916(https://doi.org/10.1042/BJ20101247) found that DNAJB2a suppresses aggregation but does not improve luciferase refolding. Independent PMID:20889486(https://pubmed.ncbi.nlm.nih.gov/20889486/) supports J-domain-dependent recovery of mutant Parkin recruitment and mitophagy-related function. This supports client-specific folding recovery, without turning antiaggregation into a refolding assay or generalizing the negative luciferase result to all clients.
The four obsolete GO:0051082 source assertions and the general transporter assertion are refined to GO:0140309. Its delivery requirement is supported by the independent primary abstract of PMID:15936278(https://doi.org/10.1016/j.cub.2005.04.058), which describes HSJ1 client recognition, aggregation prevention and proteasome sorting. The individual SOD1 and luciferase experiments supply binding/antiaggregation context; they are not falsely described as direct delivery measurements. The activity is integrated with client recognition and triage, not treated as an unrelated third core.
The two MLF1 associations are retained as non-core positive evidence. The historical self-product WITH/FROM assertion from PMID:21625540 remains unresolved at the exact assay level, but the independently verified DNAJB2a extension study reports wild-type as well as mutant polydisperse oligomers. That supports a gene-level self-association without establishing a homodimer or retroactively verifying the original pair. This follows the project’s explicit curation instructions for preserving positive generic binding when a more informative specific activity is not justified.
Two inherited removals claiming that PMID:9553041 concerns only a different gene are corrected. The full original genuinely tests a human HSJ1 J domain in engineered WT1 fusions. Its growth effects depend on the fusion context and WT1 DNA-binding function, so assigning them to native full-length DNAJB2 is marked over-annotated. The experimental observation is preserved. Likewise, constitutive expression is not a reason to deny response to unfolded protein: the older cloning paper is limited evidence, while later target studies establish action on misfolded clients.
Current HPA ICC/IF records resolve the previous access-based uncertainty about nuclear-membrane localization, without assigning the antibody signal to a specific splice product. The official products remain P25686-3/HSJ1b, 324 residues, and P25686-2/HSJ1a, 277 residues. Historical 274-residue constructs and engineered domains are distinguished from those natural products.
Four short literal publication anchors occur once each in the YAML cores. Full source access, assay/construct limits and public locators are in the source guide. Publication cache flags remain unchanged even where an original was recovered separately. The existing provider report is preserved rather than replaced with manually authored provider text.
The current normal export now occupies DNAJB2-goa.tsv, the path used by validation. The exact historical export is preserved in source-snapshots/DNAJB2-goa-before-review.tsv; the earlier current-export link remains a byte-identical copy. This supersedes the earlier paragraph's implication that the historical export remains at the standard GOA path. The source provenance records both versions and their exact assertion differences.
The historical file has 46 records representing 41 distinct term/evidence/reference triples; the current file has 45 records representing 40 triples. Six historical triples are absent from the current file: four GO:0051082 assertions, the GO:0051087 IEA assertion from GO_REF:0000117, and the GO:0140318 TAS assertion from PMID:28031292. Their zero-based review indices are 4, 9, 10, 16, 27 and 38. They are retained with retired: true, preserving their source fields and scientific MODIFY decisions. Retirement documents the source version change, not a newly inferred loss of function. The other 35 historical triples plus the five genuine additions cover all 40 current triples. No evidence code, source identifier or scientific action was changed to obtain a validation match.
The second coupled client-handling core now explicitly uses GO:0140309 unfolded protein holdase activity. The existing live GO:0140036 reader annotation remains accepted and its ubiquitin-dependent recognition mechanism remains in the core description. This expresses the independently supported carrier function even though the five historical replacement recommendations now belong to retired source contexts. No new annotation row or third core is introduced, and retirement metadata remains unchanged. DNAJB2 performs this carrier activity; a contributes_to qualifier is not used merely because other proteins participate in client handling.
The protein-folding row is refined to GO:0042026 using the independently demonstrated Parkin functional recovery. The original SOD1 source still establishes association, antiaggregation and altered turnover, not recovery of SOD1 catalytic activity. The folding-regulation and ATP-dependent-activity regulation assertions retain their distinct meanings; neither is an is_a/part_of ancestor of the other. No universal productive-refolding outcome is asserted.
The source guide now adds exact public original-section/page locators. Separate original recovery is not equivalent to changing a cached publication's availability flag. In particular, PMID:12754272 remains abstract-only; no full-original access was claimed for it. The accepted manuscript, original PMC text and original scanned pages used for other papers remain distinct from the unchanged normal caches.
The source snapshots and provenance record are retained. DNAJB2-goa.tsv is the standard current export used by the validator. DNAJB2-current-goa.tsv is the frozen copy linked in the first notes entry, recording that retrieval rather than a second independently maintained current database. The archived prior GOA records the evidence set underlying the preserved historical source assertions. The provenance JSON was generated from the actual two exports and normal-fetch receipt; its counts, set differences and hashes are reproducible calculations, not fabricated source annotations. These preserved artifacts must not be refreshed independently or interpreted as current biological truth. No archive, source object, cache, provider report or prior history is deleted.
The two preserved GOA exports now reside outside the bioinformatics folder in source-snapshots/. The prior-main snapshot retains the exact historical bytes; its date records this review's archival capture rather than an unknown original download date. The QuickGO snapshot retrieved at 20:10:55 UTC retains the exact fetched bytes and replaces the misleading DNAJB2-current-goa.tsv filename. These snapshots are immutable; only the standard DNAJB2-goa.tsv is the input regenerated by future normal fetches.
The source provenance is now YAML outside the bioinformatics folder. Its source hashes, modes, retrieval record, counts, six historical-only triples, five retrieved-only triples and retirement mappings are preserved. Only destination-path metadata and the frozen-copy explanation changed, with an additive relocation record. The historical current fields refer to the dated retrieval, not a live source service. Earlier prose describing the JSON format and old filenames records the previous layout; its three snapshot/provenance links now resolve to the new paths. No scientific review field, product, quotation, raw source byte stream, provider output or prior history was changed or lost.