Ensembl Compara donor provenance for human ADNP (Q9H2P0)

Rendered by analyze_compara_donors.py --render-snapshot from the saved results.json QuickGO/PubMed source snapshot. This rendering does not refresh network data or alter the raw snapshot.

Summary

• human GOA rows: 53
• of which projected by Ensembl Compara (GO_REF:0000107): 26

• donor entries: UniProtKB:Q9JKL8, UniProtKB:Q9Z103

• rows whose donor reference classifies as MIXED: 2

• rows whose donor reference classifies as PEPTIDE_ONLY: 13
• rows whose donor reference classifies as PROTEIN: 11

PEPTIDE_ONLY means the donor reference's title/abstract names the synthetic ADNP-derived octapeptide (NAPVSIPQ / D-NAPVSIPQ / the related ADNF-9 peptide) without a matched marker of the full-length gene product being detected or perturbed. This classifier label does not establish the absence of a native-protein experiment from the full paper.

Rows

References where restricting protein markers to ADNP-naming sentences changed the marker set: PMID:15800376, PMID:15963648, PMID:19047645. The scoping matters: on PMID:19047645 the unscoped match fired on an siRNA directed against Fyn kinase, not ADNP.

HP1-interaction motif scan

P.V.[LMIV] over ADNP_HUMAN (Q9H2P0, 1102 aa):

Expected matches under a null from the protein's own residue composition: 0.758. The match count alone does not establish an HP1 contact or determine whether mutagenesis has been performed. Sequence statistics and functional experiments are distinct evidence; consult the current gene review for the experimentally mapped interaction.

The saved JSON's historical interpretation field is not a current literature assessment. Subsequent source review documents ADNP motif and reciprocal HP1-interface perturbations in PMID:21267468 and ADNP motif mutants in PMID:38960717; see the gene review and notes. The sequence matches above are retained independently of that evidence.

Reference-projection test

For each donor reference, the number of distinct gene products it annotates anywhere in GOA. A reference annotating a complex plus every subunit with identical evidence would be a projection rather than N independent findings.

Retraction / erratum check

Checked 33 PubMed records: every donor reference above plus every PMID: entry in the review's own reference list.

None carries a RetractionIn, ErratumIn, ExpressionOfConcernIn or republication link on its own PubMed record. Negative result, recorded so the next reader knows the check was run. Note the known limitation: a correction whose own PubMed id is null is invisible to this route and would need a Crossref relation/update-to lookup.

Rat experimental terms that were NOT projected to human

The negative control on the Compara filter itself: rat Adnp terms with their own experimental evidence that did not reach human ADNP.

• GO:0032147
• GO:0050731

Limitations

• The classifier reads the donor reference's title and abstract. Most of these papers have full_text_available: false in publications/, so a peptide-only call is a strong screen, not a reading of the methods. Each call is reported with the exact markers that fired so it can be re-checked.
• MIXED means both peptide and protein markers fired; those rows are not counted as peptide-derived, which is the conservative direction.
• Entity counts come from QuickGO's fully paginated annotation search (numberOfHits == len(results) asserted), so they are distinct gene-product ids, not annotation totals.