MFF notes

Deep research: not run (falcon times out in this environment; perplexity-lite unavailable). Review based on cached publications, UniProt and PubMed.

• MFF is the main outer-membrane DRP1 receptor PMID:21149567.
• Tail-anchored in the outer membrane, also on peroxisomes PMID:18353969; PMID:23921378.
• Cytoplasmic domain is a dimer PMID:23530241.
• Core MF chosen: GO:0043495 protein-membrane adaptor activity (covers mitochondria and peroxisomes; GO:0180050 is a mitochondria-only child). Generic protein-binding rows removed; the PMID:21149567 row modified to GO:0043495.

GO:0005096 across the full human carrier set

The suggested_questions entry asserts two things about the human carriers of GO:0005096
GTPase activator activity: that none of the four mammalian outer-membrane DRP1 receptors carries
the term, and that the only self-assembling dynamin-family GTPase that does is DRP1 itself. Both
are claims about every human carrier, so they are sourced from QuickGO over the whole set rather
than from the gene folders in this repository, which hold 35 carriers out of 2958 human folders.
Query run 2026-10-14:

curl -H 'Accept: application/json' \
  'https://www.ebi.ac.uk/QuickGO/services/annotation/search?goId=GO:0005096&taxonId=9606&geneProductType=protein&qualifier=enables&limit=100&page=1'

numberOfHits 2122 over 22 pages of 100, across 1772 distinct geneProductId accessions and
486 distinct gene symbols. The two counts differ because a gene is annotated under more than
one accession, so they are not interchangeable and the review text says symbols. Every one of the
2122 results carries a symbol (zero without), which is what makes absence from the 486 symbols
equivalent to absence from the carrier set; without that, a 3.6-fold ratio would be equally
consistent with some accessions returning no symbol at all. Checks over that set, each stated at
both symbol and accession level, with accessions read from UniProt by exact gene symbol rather
than written from memory:

• the only dynamin-superfamily member is DNM1L itself. Checked against DNM1, DNM2, DNM3,
  DNM1L, OPA1, MFN1, MFN2, MX1, MX2, EHD1-EHD4, GBP1-GBP7 and ATL1-ATL3;
  only DNM1L matches, and its accession O00429 is present among the 1772.
• none of the four mammalian outer-membrane DRP1 receptors carries the term. MFF, MIEF1
  (MID51), MIEF2 (MID49) and FIS1 are all absent from the 486 symbols, and their accessions
  Q9GZY8, Q9NQG6, Q96C03 and Q9Y3D6 are all absent from the 1772 ids. L0R8F8, the
  AltMIEF1 alternative-ORF peptide that UniProt files under the MIEF1 symbol, is absent too, so
  the MIEF1 leg holds for both entries that carry that symbol.
• the filter does exclude negated rows, so "positive enables" is accurate. The only qualifier
  value anywhere in the 2122 results is enables, which settles the filter's semantics on its
  own. The gene-level control agrees: ARFGEF1-goa.tsv carries a NOT|enables GO:0005096 row by
  IDA and no positive row, and ARFGEF1 is absent from the 486 symbols with its accession
  Q9Y6D6 absent from the 1772 ids. That ties the in-corpus arithmetic (35 symbols with a row,
  34 positive) to the query.

The first version of this control cited Q9Y6D5 as ARFGEF1's accession. Q9Y6D5 is ARFGEF2
(ARFGEF2-uniprot.txt:2); ARFGEF1 is Q9Y6D6 (ARFGEF1-uniprot.txt:2, and the accession on its
own GOA row). ARFGEF2 holds no GO:0005096 row at all, so that leg was not a weak check but a
vacuous one on a gene that was never a carrier, and its negative result said nothing. Corrected
above; the conclusion was already carried by the qualifier leg.

The 35 folders in this repository are not the carrier set, which is why the claim is sourced from
the query rather than from them: ARHGAP1, ARHGAP5, ARHGAP35, RACGAP1, RABGAP1 and TSC2
are each in the QuickGO result and none has a folder here. The five dispositions the question
lists (DNM1L, ARHGAP11B, ARHGAP36, AGFG1, AGFG2) remain claims about this repository's own reviews
and are scoped to it in the question text.

What these checks do not establish, recorded because the first version of this section
overreached. Check one asks which carriers are dynamin-family GTPases; check two asks whether
DRP1's own receptors carry the term. Neither reaches the broader claim that no carrier is a
receptor or adaptor for any self-assembling dynamin-family GTPase - DNM1, DNM2, DNM3, OPA1, the
mitofusins, the EHDs and the atlastins each have their own partners, and none of them is covered
here. No further query can close that gap either, because "is a receptor or adaptor for a
dynamin-family GTPase" is not a field any annotation set returns; establishing it would mean
classifying 486 gene products by a property GO does not record. The review text therefore asserts
only the two checked claims, which is what the MFF case needs.