magi3b (LOC564220) notes

Directory name LOC564220 follows the PANTHER v19 label. Current names: ZFIN magi3b
(ZDB-GENE-060503-301, chromosome 6); NCBI Gene 564220 magi3b; Ensembl ENSDARG00000025974.
UniProt A0A8M6YYZ5 (TrEMBL, 1406 aa). Paralog: wu:fi36a10 (NCBI/RefSeq magi3a, ZFIN still
wu:fi36a10). Reviewed as part of DANRE_DUPLICATION batch 3 (random sample).

Deep research

Deep research was not available for this batch (Edison returned 402 Payment Required, and the
OpenAI key is invalid). I searched the literature myself through Europe PMC for "magi3b",
"magi3a", "magi3 AND zebrafish", "LOC564220" and "wu:fi36a10".

Literature (zebrafish)

• America et al. 2022 (PMID:35649360), the only paper that studies both copies by name.
• Y2H screen with the Gpr124 intracellular domain: PMID:35649360
• Both copies are expressed in sorted endothelial cells: PMID:35649360
• F0 crispants: PMID:35649360
• Not additive: PMID:35649360
• The paper's magi3b in situ primers match ZFIN magi3b (LOC564220) with 0 mismatches
  (file:DANRE/wu_fi36a10/wu_fi36a10-bioinformatics/RESULTS.md).
• Choksi et al. 2014 (PMID:25139857): Foxj1-target screen. A "magi3" morphant had shorter kidney
  motile cilia, and ZFIN attached this to magi3b.
  PMID:25139857
  The GFP-tagged protein was not ciliary. It was cytoplasmic/nuclear/membrane:
  PMID:25139857

Family background

• MAGI GK domains are degenerate: PMID:37163606
• The UniProt "ATP-binding" segment (118-125) has no Walker A motif (FQKGSIDH) in either copy
  (RESULTS.md). The kinase, transferase, ATP binding and nucleotide binding IEA rows are therefore removed.
• Human MAGI3 at tight junctions, with PTEN: PMID:10748157

Sequence and expression (my analyses; file:DANRE/wu_fi36a10/wu_fi36a10-bioinformatics/RESULTS.md)

• 58.2% identity to wu:fi36a10 (compare_pair.py). magi3b is the more divergent copy: 55.1% to gar
  magi3 vs 63.5% for wu:fi36a10.
• Both copies keep every domain. The C-terminal disordered tail is poorly conserved everywhere
  (about 20%).

• Bgee RNA-Seq: magi3b has 9 calls (highest in retina); wu:fi36a10 has 22 calls; every magi3b call is shared.

• Synteny (synteny_output.md): wu:fi36a10 sits next to ptpn22 and RSBN1, as gar magi3 does. The magi3b
  region shares no gar-window gene, and there is no double-conserved synteny. Ensembl Compara dates the
  split to Euteleostomi. The TGD origin is PANTHER tree-based only.

Decisions

• Kinase / transferase / ATP / nucleotide binding (UniRule IEA): REMOVE (pseudo-GK).
• Cilium IMPs: KEEP_AS_NON_CORE (single-morpholino screen).
• Localization IBA/IEA/IDA: ACCEPT. Signal transduction IBA: ACCEPT (general).
• No NEW terms. ZFIN curated only adgra2 from PMID:35649360. The magi3 data are F0 crispants, so I
  did not add a brain angiogenesis term (see "do not add what curators declined").