GEMIN5 selects RNA substrates for snRNP assembly and regulates translation. Four direct molecular activities are represented: snRNA recognition, cap recognition, SMN-mRNA 3′-UTR binding and ribosome binding. Cap binding does not require obligatory SMN membership; ribosome binding does not imply an integral ribosomal subunit or a proven elongation-stalling mechanism. Positive SMN-specific translation and broader repression remain separated by experimental context.
The 73 incoming annotation contexts are retained. Sixteen generic protein-binding terms are removed while their observed partner pairs remain recorded. Six refinements specify mRNA binding, viral-IRES 5′-UTR binding, DEAD-box helicase binding, eIF4E binding or gems. Broad splicing, complex assembly and nuclear-body assertions are represented by supported, more specific claims. No new annotation rows or natural products are introduced.
Selected primary evidence links original articles and identifies the relevant experiments. Source evidence records cache byte offsets, official UniProt JSON fields and the separate thesis PDF identity. Partner classifications use FUNCTION, SUBUNIT or SIMILARITY comments, rather than recommended protein names. Repeated citations share these anchors; a reference-only citation is not a newly checked verbatim quotation. Generic UniProtKB excerpts were matched independently to official records; the local-file record-section backfill does not process those identifiers.
The original review contained 194 quote associations and four core entries. Its full Git object and all 194 pointer/hash associations are archived. The first replacement used five selected excerpts and combined the cores into two entries. This amendment expands the selected excerpts and restores explicit representation of all four direct activities. Current excerpts are limited to 25 words per source in aggregate, including repeated placements. Historical hashes are provenance, not additional current quotations or a claim of fresh verification.
The membrane assessment retains two positive identifications from separately accessed author-thesis Tables S7/S8. The printed/PDF pages, protein identifier and PDF hash are supplied. These data support fraction association, not integral insertion, topology, enrichment or an established fractionation artifact. The final journal supplement remains unavailable; it is not replaced by relabeling the thesis.
The mRNA-interactome row concerns poly(A)-RNA capture and is refined to mRNA binding independently of the SMN 3′-UTR result. The viral-IRES fragment row is refined to mRNA 5′-UTR binding using the original substrate study and official IRES identity. The full 462-nucleotide RNAcentral record remains distinct from the shorter reagents. Broad HPA nuclear-body localization is redundant with the supported gem location, while the direct imaging row identifies gems specifically. No unexamined HPA focus is relabeled. The source-specific nuclear-body distinction remains disclosed. The rules-based consistency checker supports row exceptions, but its current table has no SAME_TERM TOO_GENERAL/MODIFY rule; an exception for this pair would be stale. The separate legacy checker produces the nuclear-body advisory and has no exception mechanism.
There is one natural 1,508-residue product. Experimental fragments, immunogens and variant constructs are not additional isoforms. The original 73 rows produce 71 native keys; two genuine source contexts are restored separately, with dates and identifiers preserved. The cytoplasm extension is retained. Six histories and the original 7,315-byte notes prefix remain preserved.
The 44 parent source files are unchanged. PMID:23221641 is the sole addition, acquired as a full XML cache with the approved title and access assessment. Four PAINT assertions follow the verified PTHR46362 ancestry; the target among donors is legitimate experimental grounding. Reactome input/output chains distinguish cytosolic and nucleoplasmic GEMIN5 from event locations and other proteins’ catalytic roles. The source ledger retains these checks and the original native provenance. Amendment default validation passed with one nuclear-body advisory; strict GOA matching passed for all 73 originals, and status assessment confirms DRAFT. A genuine EDIT history, independent excerpt replay and private rendering accompany the candidate. The earlier validation section records the published parent.
There are now 27 selected excerpt placements: 16 resolve to repository publications or the local manual source account, and 11 resolve to the retained official UniProt JSON fields documented above. Repeated excerpts count toward the same 25-word source limit. The added local membrane excerpt is a table row from the manual account of the separately inspected thesis, not a quotation from the inaccessible journal supplement. The broad HPA location remains separate from the direct gem experiment.
This narrow excerpt amendment passed default validation with the same legacy nuclear-body advisory, strict GOA matching and status assessment. Its genuine EDIT history and private render are included, and all 27 excerpt placements and aggregate budgets passed independent replay. Earlier completed outcomes describe the published parent. No source acquisition, biological decision, product or core-function change was made.