ARFGEF1 bioinformatics results

Generated by make_results.py from the committed TSV/JSON outputs. Do not
hand-edit — re-running regenerates this file. Reproduce with the commands in
README.md; verify with make_results.py --check.

1. WITH/FROM resolution and donor evidence

Every WITH/FROM token in ARFGEF1-goa.tsv resolves: 0 unresolved.
Of 59 distinct tokens, 42 name a protein — 39 Swiss-Prot, 3 TrEMBL. The rest are PANTHER ancestral nodes,
InterPro signatures, ARBA rules, UniProt subcellular-location ids and an Ensembl
protein id, none of which is a gene product.

Donor evidence was then queried per (donor, propagated term) pair against QuickGO
with goUsage=descendants. 43 of 44 (donor, term) pairs carry the donor's own experimental
evidence for the propagated term. The single exception is Q86KG9 (Dictyostelium), which holds
GO:0016192 by IBA only. No propagated row on this gene can be argued down on the
grounds that its sources carry only the same family-level inference.

IBA row protein donors with own experimental evidence Swiss-Prot / TrEMBL
GO:0005085 guanyl-nucleotide exchange factor activity 18 18 16 / 2
GO:0016192 vesicle-mediated transport 18 17 16 / 2
GO:0005802 trans-Golgi network 3 3 —

For GO:0005085, all 18 donors hold only GO:0005085 — not one holds a descendant.

All 18 protein donors of the GO:0005085 IBA carry their own experimental evidence for that term, 16 of them as Swiss-Prot entries, and none holds a descendant of it.

For GO:0016192, by contrast, the donors' own terms are spread across 13 distinct descendant processes (GO:0006887, GO:0006888, GO:0006890, GO:0006891, GO:0006892, GO:0006893, GO:0006895, GO:0016197, GO:0032509, GO:0042147, GO:0043001, GO:0048193, GO:0048205), so the generic parent is the
genuine least common ancestor of a heterogeneous donor set rather than an
under-specified term.

For GO:0005802, every donor holds GO:0005802 itself, so the propagation lands at the same
specificity as its donors rather than above them.

All 3 protein donors of the GO:0005802 IBA carry their own experimental evidence for GO:0005802 itself rather than for a parent of it.

2. GO:0005086 and the substrate-specific GEF terms are gone

QuickGO resolves GO:0005086 to guanyl-nucleotide exchange factor activity (GO:0005085), and
GO:0005085 now lists 14 merged ids among its secondaryIds and has no
substrate-specific is_a children. The merged ids are:

GO:0005086 GO:0005087 GO:0005088 GO:0005089 GO:0005090 GO:0008321 GO:0008433 GO:0016219 GO:0016220 GO:0017034 GO:0017112 GO:0017132 GO:0019839 GO:0030676

The merged ids themselves now resolve to the parent, so their old labels are
gone from the current ontology. What survives is the synonym list: GO:0005085
carries 10 narrow synonyms naming substrate-specific exchange
activities, which is the retrievable evidence of what was absorbed:

Its only children are 2, neither of them a substrate-specific activity:

So GO:0005085 is already maximal for ARFGEF1, whose defining property is that it
is an ARF1/ARF3 exchange factor. The substrate can only be recorded as an
annotation extension (RO:0002233 has_input) or in core_functions[].substrates.

The same has happened on the GTPase-binding side, which matters because this
review proposes GO:0031267 small GTPase binding for the ARL1 interaction:
11 ids are merged into GO:0031267, and its narrow synonyms include
'ADP-ribosylation factor binding', 'ARF binding' — so an Arf-specific
GTPase-binding term did once exist and is now only a synonym. There is no
Arf-family child to propose.

3. A paralog shadow set on MYO9A from PMID:15644318

Querying GOA by reference rather than by gene returns 18 annotations for
PMID:15644318. MYO9A has acquired an exact copy of the MYO9B partner set (ARFGEF1, RHOA) but none of the MYO9B functional rows —
MYO9B additionally holds GO:0005096, GO:0032011, which MYO9A does not.

gene product GO term qualifier evidence with/from
MYO9A B2RTY4 GO:0005515 enables IPI P61586
MYO9A B2RTY4 GO:0005515 enables IPI Q9Y6D6
RHOA P61586 GO:0005515 enables IPI B2RTY4
RHOA P61586 GO:0017022 enables IPI Q13459
MYO9B Q13459 GO:0032011 NOT involved_in IDA
MYO9B Q13459 GO:0005096 enables IDA —
MYO9B Q13459 GO:0005515 enables IPI P61586
MYO9B Q13459 GO:0005515 enables IPI Q9Y6D6
ARFGEF1 Q9Y6D6 GO:0005096 NOT enables IDA
ARFGEF1 Q9Y6D6 GO:0005085 enables IDA —
ARFGEF1 Q9Y6D6 GO:0005515 enables IPI Q13459-2
ARFGEF1 Q9Y6D6 GO:0017022 enables IPI B2RTY4
ARFGEF1 Q9Y6D6 GO:0017022 enables IPI Q13459
ARFGEF1 Q9Y6D6 GO:0034260 involved_in IDA —

MYO9A partner set: ARFGEF1, RHOA. MYO9B partner set: ARFGEF1, RHOA. Identical: True.
The paper's title, abstract and every described experiment concern myosin IXb. The
full text could not be obtained (JBC returns HTTP 403), so the affected review row
is left UNDECIDED rather than removed.

4. Interaction partners: identity and experiment independence

All 12 GO:0005515 IPI partner accessions resolve to reviewed
(Swiss-Prot) canonical entries with the canonical length — no TrEMBL fragments, no
partial ORFeome clones.

partner accession len records pubs distinct experiments spoke-expanded max MI partner's own IntAct partners
NCL P19338 710 5 1 1 3 0.5 303
NCKAP1L P55160 1127 0 0 0 0 — 7
MYO9B Q13459-2 2157 2 1 2 0 0.56 79
PDE3A Q14432 1141 6 1 2 5 0.58 59
KANK1 Q14678 1352 8 1 2 3 0.63 35
KIF21A Q7Z4S6 1674 11 2 6 3 0.71 58
KIF21A Q7Z4S6-2 1674 11 2 6 3 0.71 58
Dpy30 Q8K3E7 99 0 0 0 0 — 1
TBC1D22A Q8WUA7 517 0 0 0 0 — 45
MYCBP Q99417 103 7 4 6 4 0.79 141
TBC1D22B Q9NU19 505 1 1 1 1 0.35 175
ARFGEF2 Q9Y6D5 1785 14 4 6 8 0.84 99

Counting distinct (publication, detection-method) pairs rather than UniProt's
NbExp did not collapse any partner to one screen logged several ways.

5. Reference scope — no complex-projection signature

12 of the 18 references supporting ARFGEF1 annotations are
gene-focused, annotating four entities or fewer across all of GOA. None shows the
pattern of a complex-level phenotype distributed to every subunit.

3 references are too large to enumerate from one page (PMID:33961781, PMID:35271311, PMID:40205054); their entity
counts are reported as unavailable rather than derived from a partial page. All
three are proteome-scale interaction maps, and every row they contribute to
ARFGEF1 is GO:0005515 — what they contribute to the rest of GOA was not
enumerated and is not claimed here.

6. Literature coverage

Union of the affinage citation block and every RX PubMed= line in the UniProt
entry: 48 PMIDs. Restricted to the 26 functional papers affinage
returned:

bucket n PMIDs
annotate ARFGEF1 6 14973189, 15644318, 17227842, 18292223, 19332778, 20360857
annotate only other genes 3 22291037, 23386609, 24090963
no GO annotation anywhere 17 10393931, 12606707, 16467138, 17360629, 17640864, 18417613, 23220274, 23918382, 24198228, 27162341, 27834853, 28414797, 29740613, 31678406, 32415087, 35090882, 36562883

That is, of the 26 functional papers affinage returned, only 6 have
produced any GO annotation on human ARFGEF1, and 17 have produced none
anywhere in GOA — including PMID:10393931, which measured the exchange reaction
and the brefeldin A inhibition constant on the human protein.