ABI1 re-review, 2026-09-11

Reviewed all 56 existing annotation rows against the cached publications, UniProt and existing Falcon research. PubMed and full-text searches were used for ambiguous claims. Publication caching recovered 41/41 existing references. New Falcon research with perplexity-lite fallback was launched; provider status is recorded below when complete.

The phosphatase/ABA mechanism is grounded in direct literature rather than synthetic descriptions of research or PANTHER files. [PMID:8898906, "Mg2+ or Mn2+ dependence"] [PMID:19924127, "the SnRK2 kinases are kept inactive by the PP2Cs through physical interaction and dephosphorylation."] SnRK1 and MAPKKK18 extend the signaling substrate scope [PMID:24179127; PMID:26443375].

The REMOVE decision for PMID:17267444 was withdrawn: an abstract about SCAR/WAVE does not establish a gene-name mapping error. Search results point to divergent ABI/ABIL naming, but assay construct identity remains unverified. The row is UNDECIDED. This also applies to interaction-map rows whose specific interaction tables were unavailable.

The 1991 cold study reports "Cold-regulated expression of all three cor genes, however, was nearly the same in wild-type and abi1 mutant plants." PMID:1834244. Its row is UNDECIDED pending full context; the independent 1995 cold-response result remains non-core PMID:12228349.

Early calcium-binding assertions were sequence predictions [PMID:7910981; PMID:8197457]. Later experiments question physiological EF-hand regulation PMID:8898906. The calcium annotations remain over-annotated with an evidence-based rationale. Broad cation binding is retained non-core because magnesium/manganese dependence is experimentally established. The broad PAINT signaling-regulation annotation is accepted; target-specific ABA evidence does not invalidate an ancestral broad function.

Specific citations and snippets were added only where the accessible text supports the assertion. IEA mappings use cited rationale without mandatory snippets. Several abstracts do not expose the ABI1-specific assays; those rows retain explicit uncertainty rather than invented supporting quotations.

Falcon research refresh completed successfully and generated an updated report plus provider artifacts. Its synthesis confirms the PP2C/SnRK2 mechanism and flags kinase-side osmotic regulation as context; no unverified new experimental annotations were added from that synthesis. The refresh of PMID:24179127 recovered full text confirming ABI1-SnRK1.1 interaction and phosphatase activity. Proposed replacement GO identifiers were checked with QuickGO and match current labels. Removed the obsolete failure-classification block from the now-accepted PAINT annotation instead of retaining an unsupported evolutionary defect claim.

2026-09-12: PR2996 review follow-up

Applied the canonical annotation-reviewer skill to the 17 remaining generic protein-binding over-annotation decisions: 13 become REMOVE as uninformative labels with interaction evidence preserved; three become MODIFY where the cited source supports an actual activity. PMID:19805022 and PMID:19924127 support serine/threonine phosphatase activity [PMID:19805022, "efficiently inactivated ABA-activated SnRK2s via dephosphorylation of multiple Ser/Thr residues in the activation loop."]; PMID:26443375 supports protein kinase inhibitor activity [PMID:26443375, "Notably, we identified abscisic acid insensitive 1 (ABI1) PP2C as a MKKK18-interacting protein, and demonstrated that ABI1 inhibited its activity."]. GO:0004722 and GO:0004860 names and definitions were verified through official QuickGO. Kinase inhibition does not imply a phosphatase-independent mechanism.

PMID:19955427 becomes UNDECIDED because its cached abstract describes PP2CA-OST1-SLAC1 but does not expose the ABI1-specific experiment. This does not establish misattribution. The previously unresolved generic-binding rows stay unresolved under the skill's evidence-access exception; the reviewer's recommendation to restore MARK_AS_OVER_ANNOTATED was not adopted because the canonical skill explicitly disallows that action for generic binding. Replaced all 11 repeated UNDECIDED summaries with source- and term-specific explanations. Source identities and physical-interaction quotations are unchanged. No new research job was needed for these bounded review corrections.