The source set contains sixteen annotations for human ASXL2 (Q76L83) and two recorded protein products. The source annotations and product definitions are preserved. UniProt describes isoform 2 as lacking residues 1–260 and 838–1094 of the displayed sequence; the general GOA rows do not establish isoform-specific functions.
The initial synthesis distinguishes two molecular roles: BAP1 activation in PR-DUB and transcriptional coactivation. The primary ASXL2 study reports a regulatory contribution to ubiquitin recognition and BAP1 activity; BAP1 performs hydrolysis. The independent DEUBAD study includes ASXL2 activation assays despite naming ASXL1 in its title. These support candidate activator annotations rather than assigning ASXL2 intrinsic deubiquitinase activity. Root read both indexed original abstracts and identifiers; the independent annotation reviewer additionally read selected construct/assay passages and the publisher Figure 1 activation experiment. Complete Methods and all panels have not yet been read. Sources: PMID:26416890(https://pmc.ncbi.nlm.nih.gov/articles/PMC4661380/), PMID:26739236(https://pmc.ncbi.nlm.nih.gov/articles/PMC4729829/).
The receptor study tests human ASXL2 in human HEK293 cells and in mouse 3T3-L1 differentiation experiments. The independent reviewer read Experimental Procedures, Results and Figures 1–6 legends. These support PPAR association and transcriptional activation; adipogenesis remains a specific biological context, not a separate core activity. Endogenous ASXL proteins were co-immunoprecipitated with transfected FLAG-PPARgamma, so that assay must not be described as an entirely endogenous interaction. Root's final literal citations await the normal publication cache. Source: PMID:21047783(https://pmc.ncbi.nlm.nih.gov/articles/PMC3020743/).
The DNA-binding IEA specifically names IPR026905, the ASX-like PHD family. The HARE-HTH domain is the separate IPR007759 record supporting the transcription-regulation IEA. Chromatin association alone does not resolve a direct DNA-binding claim. Recent isolated-PHD experiments constrain a histone-tail-reader interpretation and identify MBD5/6 interactions; they do not test direct DNA recognition. Root read the official abstract and identifiers, and the independent reviewer read exposed figure legends, with full Methods and raw data still unexamined. Source: PMID:40729547(https://pubmed.ncbi.nlm.nih.gov/40729547/).
The SPOP interaction remains source-specific uncertainty. UniProt positively records SPOP (O43791) with NbExp=6. That count does not establish which wild-type/mutant experiment supports the annotation, and the existing cache does not by itself show that the exact pair data were examined. No negative binding conclusion is warranted. Source: PMID:35512704(https://pmc.ncbi.nlm.nih.gov/articles/PMC9597701/).
The independent reviewer traversed both Reactome event hierarchies to the human Q76L83 ASXL2 member and read the HPA nucleoplasmic antibody/cell-line table. Those database observations support localization and complex context without implying review of every underlying experiment. Sources: HPA ASXL2, Reactome BAP1 assembly, Reactome PR-DUB reaction.
The prospective decisions are eleven ACCEPT, three KEEP_AS_NON_CORE and two UNDECIDED, with two candidate NEW molecular functions and no new biological process. Positive target evidence in PAINT WITH/FROM is legitimate; no adverse conclusion is drawn from donor count or paralog identity. The detailed ancestral placements have not been reconstructed.
The ordinary research-provider attempt failed without producing a report. One normal request each for publications 26416890, 26739236 and 40729547 failed without creating files. Existing source files were preserved. The already queued shared source request owns 21047783, 30664650 and 36180891; duplicate requests were avoided. The YAML remains a source stub while publication recovery and final citation checks continue.
Normal source recovery is complete. All seven required papers and both pathway summaries are available; the existing caches were preserved. The final review keeps all sixteen original source objects and both products: eleven ACCEPT, three KEEP_AS_NON_CORE and two UNDECIDED, plus two NEW molecular functions. The proposed activities are BAP1 activation and transcriptional coactivation. There is no NEW biological process, intrinsic deubiquitinase claim or isoform-specific activity claim.
Root independently read the actual XML activation Results and cloning/protein-expression Methods in PMID:26739236(https://pmc.ncbi.nlm.nih.gov/articles/PMC4729829/). Human ASXL2 Q76L83 DEUBAD residues261-380, expressed in bacteria, activates separately produced BAP1 on H2A-K119ub. The NEW IDA activator row uses this directly inspected source as its primary reference, with PMID:26416890 corroboration; this updates the provisional primary citation without changing any original GOA source object. Complete kinetic supplements and figure pixels were not inspected.
The actual XML PMID:40729547(https://pmc.ncbi.nlm.nih.gov/articles/PMC12713335/) binding Results distinguish the ASXL2 histone-tail assays from MBD5/6 association. ITC and repeated ELISAs do not show consistent recognition of the tested histone-H3 peptides under those conditions. This does not test DNA recognition and therefore does not resolve the existing PHD-derived DNA-binding IEA. Recombinant PHD-MBD6 co-purification, NMR, zinc-binding assays, native MS and interface mutants support complex formation; the proposed atomic interface is AlphaFold-modelled. Histone assays use ASXL2 residues1375-1435; partner-association legends/Results use1399-1435, while the SUMO construct Methods list1395-1435. That boundary discrepancy is retained. Neither raw supplemental data nor figure pixels were inspected.
Root also read the complete normal abstracts for PMID:21047783, PMID:26416890, PMID:30664650 and PMID:36180891, and indexed original human-ASXL2 construct, HEK293 interaction and luciferase Methods for PMID:21047783. The coactivator inference is based on receptor association and transcriptional output. Mouse 3T3-L1 differentiation remains contextual. The coactivator definition and deubiquitinase-activator definition support distinct MF roles, separate from the seeded receptor-binding/chromatin-binding terms and complex membership.
Six literal cached findings support the synthesis. Cache availability is recorded independently of actual article-reading scope: PMID:21047783 and PMID:26416890 remain abstract-only even where original indexed Methods/Results were read externally. The SPOP edge stays UNDECIDED, with the positive curated interaction retained and no false negative conclusion. The two core entries describe regulatory work and transcriptional coactivation without treating the complex name as evidence for universal gene repression.
The source-status blocks now distinguish independently supported vocabulary/experimental sources from unexamined ancestral placements. SL-0191 supports the nuclear mapping; ASXL2 self-WITH/FROM entries are experimental descendant evidence, and ASXL1 has directly supported PR-DUB membership. Unexamined donor entries are omitted from reviewer metadata while every original WITH/FROM source remains unchanged. PTN-specific uncertainty is separate from independently corroborated target biology. InterPro comments no longer refer to PAINT.
The official GOA mapping documentation identifies SL-0191 as Nucleus mapped to GO:0005634. Indexed original PMID:21047783 Methods and figures distinguish tagged human ASXL2 in mouse 3T3-L1 promoter ChIP from endogenous differentiation time-course assays. The description of those experiments now specifies both protein and cellular context. The canonical cache remains abstract-only.
Chromatin association and PHD-dependent MBD5/6 assembly are explicit within the existing PR-DUB core, supported by PMID:36180891 and PMID:40729547. Both core entries use nucleoplasm. PMID:30664650 adds complex-level corroboration to coactivation while PMID:21047783 remains its direct receptor-assay source. Directly checked references carry explicit VERIFIED assessments within their documented reading scope.
The ASXL2-SPOP row remains UNDECIDED. PMID:35512704 names SPOP F133L, but that does not establish the genotype of the uninspected ASXL2 pair in Data S1. The project action definition reserves REMOVE for an assertion judged unlikely to be correct; generic informativeness alone does not establish that judgment here. The term-specific skill also retains UNDECIDED when the relevant evidence cannot be adjudicated. No false interaction or mutant-specific assay is invented to resolve this policy disagreement. The PHD DNA-binding row likewise remains unresolved because the cited protein/histone assays do not test direct DNA recognition.
All sixteen original assertions, both existing NEW molecular functions, both products and the action tally are unchanged. No new biological process, source-cache modification or unsupported negative evidence was introduced.