awd review notes

All four seeded annotations and the original function paragraph are supported by the intact same-gene enzyme core. The selected 168 aa PC/PD sequence has 15 extra N-terminal residues, while the reviewed record questions that initiation. This discrepancy is documented without alleging a prediction-input error. PMID:1320004(https://pubmed.ncbi.nlm.nih.gov/1320004/) purified the fly enzyme and PMID:7559441(https://pubmed.ncbi.nlm.nih.gov/7559441/) connects conserved-site mutations to catalytic impairment. Native enzyme assays plus exact core conservation support the transfer.

Provider research assessment

The completed Falcon synthesis was inspected. It supports canonical NDP kinase chemistry and identifies endocytic trafficking as a well-established cellular role. The primary PMID:24528630 was cached with full text and read: live trafficking, compartment markers, mutant clones and epistasis place Awd in early-endosome maturation needed for Notch progression. This supports a NEW early-to-late-endosomal transport annotation, with ISS specifying transfer to the exact longer PC/PD product. Awd is not the receptor protease, and local GTP channeling to Rab5/dynamin is not directly established. The provider’s His119 refers to the 153-residue form; it maps to His134 in the exact accession. The extracellular-release thesis observations and disputed protein-kinase activities do not change the core NDP-kinase assessment. PMID:11768313 is a review, not a purified-protein primary experiment.

2026-10-20 — Full re-review, independent of the ProtNLM2 predictions

ProtNLM2 files were not opened or used.

Identity of the benchmark accession

• A0A0B4LHX6 is the FlyBase awd-PC/awd-PD product (168 aa; transcripts awd-RC, awd-RD). The reviewed, canonical product is P08879 = awd-PE (153 aa; awd-RE). It is an alternative gene model (initiation at an upstream ATG) of the same gene, not a fragment. The full P08879 sequence starts at target residue 16. The extension MLGTILAFFSVISAT (15 aa) is 67% hydrophobic, and UniProt's automatic SignalP feature calls 1-17 a signal peptide. UniProt P08879 flags AAF57188.3 (the 168-aa translation) as "Erroneous initiation".
• Biochemical evidence favours the 153-aa start for the native enzyme: PMID:12099695. My new mass comparison (awd-bioinformatics/mass_rereview.py, RESULTS-rereview.md) fits only Met-removed, acetylated P08879 2-153 (17080.5 vs 17082.3 Da observed). The full 168-aa form is +1640 Da off and a signal-peptide-cleaved 18-168 form is -115 Da off. Also PMID:7559441, i.e. the downstream Met is needed for protein accumulation.
• The catalytic core is fully intact in A0A0B4LHX6 (His119 -> His134; all binding residues; RESULTS.md). Function annotations are therefore correct for this product if it is made, but the description now states that it is probably not the bulk native form.

Evolutionary placement

• PANTHER 19.0 classification: awd (P08879) is in PTHR11349:SF91 "NUCLEOSIDE DIPHOSPHATE KINASE", together with yeast YNK1 (P36010) and C. elegans ndk-1 (Q93576). Human NME1 (SF69) and NME2 (SF57) are in sister subfamilies. Fly-specific evidence on the gene itself (purified enzyme) supports the NDPK activity, so no cross-species transfer is needed for the MF. The gene's PAINT IBA (FlyBase GO table) is nucleoside triphosphate biosynthetic process, from PTN000887110.
• Family review interpro/panther/PTHR11349/PTHR11349-review.yaml does not assign subfamilies and asserts no PTN node, so there is no subfamily disagreement to report. Its GO:0004550 scope is UNRESOLVED only because of the NDK-like branches (SF97 etc.). awd is not in those branches.

Annotation decisions

• GO:0004550 NDPK activity (IEA): ACCEPT (core).
• GO:0006183 GTP / GO:0006228 UTP / GO:0006241 CTP biosynthetic process (IEA InterPro): ACCEPT. Awd catalyses the terminal NDP->NTP step, so this is participation.
• Withdrawn: the earlier review's NEW GO:0045022 early endosome to late endosome transport (ISS, PMID:24528630). I re-read the full text. The evidence is necessity only: mutant clones, marker co-localisation, and Rab5CA failing to rescue ("awd function is downstream of or is required for Rab5 function"). The authors do not identify a step Awd performs; GTP supply to Rab5/dynamin remains a hypothesis (also stated in the falcon report). FlyBase curated awd's tracheal, adherens-junction and polarity phenotypes but has not given it an endosomal-transport term. Under the CLAUDE.md NEW bar (name the entity that performs the step), the step is done by Rab5/ESCRT/dynamin machinery. The question is moved to suggested_questions and suggested_experiments (a catalytic-dead rescue). In any case a gene-level process claim belongs on the canonical P08879 product rather than on this alternative model.
• core_functions: NDPK activity; GTP/UTP biosynthesis; cytoplasm (curated UniProt, PMID:2175255).

Relative and modules

• Relative: the canonical same-gene product P08879 (awd-PE). It has no review (genes/DROME/awd is keyed to A0A0B4LHX6), so it needs a review at genes/DROME/awd__P08879. Gene-level experimental GO annotations (microtubule binding, tracheal migration, adherens junction organization, etc.) sit on P08879 and should be reviewed there.
• Modules using PTHR11349 NDK steps (awd is not listed as a representative member): modules/purine_nucleotide_interconversion.yaml, modules/de_novo_pyrimidine_synthesis.yaml, modules/dntp_de_novo_synthesis.yaml; also related: modules/thymidylate_synthesis.yaml (NDPK final step, not modelled).