Are AARD's reported interactors a coherent partner set?

Generated by analyze_partners.py. Partner accessions are read from the gene's own
AARD-goa.tsv; all protein data is fetched from the UniProt REST API at run time.

Question

AARD's entire GO record consists of protein binding annotations from a single
yeast two-hybrid publication. This tests one specific, falsifiable explanation for
that set: coiled-coil bias. Sticky or self-activating preys in Y2H are enriched
for coiled-coil proteins, which associate promiscuously via heptad-repeat surfaces.
If the partners come from unrelated compartments but share coiled-coil
architecture, the assay explains the set better than any function does.

Result

Partner accessions in the GOA file 15
Successfully retrieved from UniProt 15
With a coiled-coil FEATURE 7
With a coiled-coil feature or keyword 9
Proportion coiled-coil 47%

Is that rate actually unusual? (null model)

Background, fetched from UniProt: 2059 of 20431 reviewed human
proteins carry a coiled-coil feature = 10.1%.

Both figures use the coiled-coil FEATURE, on numerator and denominator alike. For completeness, on the more liberal feature-or-keyword criterion (9/15 partners), the matching background is 10.6% (keyword:KW-0175), giving 5.6-fold and P = 4.79e-06. Mixing the two criteria - a keyword numerator against a feature denominator - would overstate the enrichment, so it is not done here.

The enrichment is therefore real rather than assumed, though with n=15 this is
a descriptive statistic and not a controlled test: the relevant comparison would
be against other single-publication prey sets from the same screen, which this
analysis does not attempt.

Partners

Accession Gene Coiled-coil segments Subcellular location
Q4V328 GRIPAP1 4 Early endosome membrane; Recycling endosome membrane; Cell projection, axon; Cell projection, dendrite; Synapse
Q86XR8-3 CEP57 2 Nucleus; Cytoplasm; Cytoplasm, cytoskeleton, microtubule organizing center, centrosome
Q7L2Z9 CENPQ 1 Nucleus; Chromosome, centromere
Q2KHM9 KIAA0753 1 Cytoplasm, cytoskeleton, microtubule organizing center, centrosome, centriole; Cytoplasm, cytoskeleton, microtubule organizing center, centrosome, centriolar satellite; Cytoplasm, cytoskeleton, microtubule organizing center, centrosome
Q16623 STX1A 1 Cytoplasmic vesicle, secretory vesicle, synaptic vesicle membrane; Synapse, synaptosome; Cell membrane; Secreted
P32856-2 STX2 1 Membrane
Q13190 STX5 1 Endoplasmic reticulum-Golgi intermediate compartment membrane; Golgi apparatus membrane
Q2M2I5 KRT24 keyword only —
Q7Z3Y8 KRT27 keyword only Cytoplasm
P61968 LMO4 0 —
O15481 MAGEB4 0 Cytoplasm
Q96HA8 NTAQ1 0 Cytoplasm, cytosol; Nucleus
Q9UBB9 TFIP11 0 Cytoplasm; Nucleus
Q9BZW7 TSGA10 0 Cytoplasm; Cytoplasm, cytoskeleton, microtubule organizing center, centrosome, centriole
A5D8V6 VPS37C 0 Late endosome membrane

Compartment spread

A genuine partner set usually concentrates in one or two compartments. UniProt
location strings are hierarchical, so they are collapsed here to their leading
compartment - 20 raw strings reduce to
14 top-level compartments:

The family offers no help either

AARD belongs to PANTHER PTHR32289 (FAM167 family; InterPro IPR051771).
For a gene this dark, paralogs are the most tractable remaining inference route,
so the family was enumerated:

Accession Entry Organism FUNCTION
Q0V7M8 F167A_BOVIN Bos taurus —
Q5RFZ7 F167A_DANRE Danio rerio —
Q9BTA0 F167B_HUMAN Homo sapiens —
Q4LEZ3 AARD_HUMAN Homo sapiens —
Q96KS9 F167A_HUMAN Homo sapiens —
A1L168 CT202_HUMAN Homo sapiens —
Q811W1 AARD_MOUSE Mus musculus —
P17257 F167B_MOUSE Mus musculus —
Q6P1G6 F167A_MOUSE Mus musculus —
Q91ZF7 AARD_RAT Rattus norvegicus —

No member of the FAM167 family has a described function in any organism.
Guilt-by-association across paralogs is therefore unavailable for AARD - not
because it lacks a family, but because the whole family is uncharacterised.
This is consistent with UniProt's PAN-GO line for AARD, which records
0 GO annotations based on evolutionary models.