ADISSP bioinformatics results

Generated by analyze_adissp.py (stdlib only; python3 analyze_adissp.py). Every number
below is computed at run time from UniProt and IntAct - nothing here is hardcoded.

1. Human/mouse identity, which licenses the ISS and Compara rows

Identity: 157/174 = 90.2%, ungapped (equal lengths, no indels).

Ensembl Compara (GO_REF:0000107) requires >= 40.0% peptide identity between orthologs: met.

SIGNAL peptide features: Q9GZN8 has 0, Q9D1K7 has 0. Neither orthologue has one, so the leaderless topology is a shared property of the pair rather than an assumption carried across.

Comparator controls (a broken comparator must not be able to report the number above): self-identity 100.0%; comparison of unequal-length sequences refused (lengths differ), as required.

2. The ADISSP-PP1 interaction in IntAct, with its null

ADISSP has 27 IntAct records. 26 of them are interactions of the ADISSP protein.

1 record(s) involve the locus but not the protein and are excluded from the partner set rather than counted as partners: mrna_adissp / (human) hsamir320a3p by clash.

Distinct protein partners of ADISSP: 13 - ARHGEF7, EPB41L5, FAM133A, FERMT2, P0C6X7-PRO_0000037312, PCDHGA9, PHKB, PPP1CA, PPP1CB, PPP1CC, PPP1R7, RALYL, TMEM69.

PP1-module partners: 4 of 13 protein partners - PPP1CA, PPP1CB, PPP1CC, PPP1R7.

Independent experimental publications recovering ADISSP with a PP1-module protein: 7 - PMID:24366813, PMID:27173435, PMID:27880917, PMID:28330616, PMID:28514442, PMID:33961781, PMID:40205054.

Methods: anti tag coip, pull down, tap. 3 record(s) were excluded as computational inference rather than experiment (socioaffinity inference); the count is stated rather than dropped silently.

The GOA annotation's own reference is PMID:32024300. Is it among the IntAct publications above? no - so these datasets are additional to, not the same as, the evidence the annotation cites.

The null: PP1 catalytic subunits are affinity-proteomics hubs

This is why the publication count above must not be read as strong replication on its own:
a protein recovered in a thousand IntAct records will reappear in many tag pulldowns. The
informative comparison is the subject-centric one - that PP1-module proteins are
4 of ADISSP's 13 distinct protein partners, so the
PP1 module dominates this small protein's own sparse interactome rather than ADISSP being one
more name on PP1's long list.

IntAct also recovers all three PP1 catalytic subunits, which is why the GO term is correctly
left at the isoform-agnostic GO:0008157 protein phosphatase 1 binding: the cited paper
identified only "PP1c" by mass spectrometry and did not resolve which subunit.

What this analysis does not show

Every IntAct record here is an affinity or two-hybrid method with at least one tagged or
overexpressed partner. None of them measures affinity, stoichiometry or an endogenous
complex, and none is in adipocytes. So this establishes that the interaction is reproducibly
detected across independent datasets; it does not establish that it is physiologically
engaged in the tissue where ADISSP's hormonal function was characterised.