IntAct curates 47 interactions against engineered MLL fusion proteins, and every accession-keyed query in this project is blind to all of them - while its PSICQUIC mirror stops at 2024

[
  "IntAct REST ws/interactor/findInteractor and ws/interaction/findInteractions, retrieved 2026-09-22",
  "IntAct PSICQUIC search/query/<accession>?format=tab25&maxResults=4000, retrieved 2026-09-22",
  "analyses/core-complex-interactome-gap/run.py (read, and executed against both endpoints)",
  "NCBI E-utilities efetch for every publication identifier quoted"
]

{
  "interactors": [
    {
      "intact_ac": "EBI-2890930",
      "name": "mll_af9_human_protein",
      "description": "MLL-AF9 fusion protein",
      "interaction_count": 32,
      "xrefs": "empty",
      "preferred_identifier": "its own IntAct AC",
      "enumerated_here": false
    },
    {
      "intact_ac": "EBI-2620068",
      "name": "mll_enl_human_protein",
      "description": "Fusion of human MLL (N-terminal region) with human ENL C-terminal residues 373-559",
      "interaction_count": 8,
      "xrefs": "empty",
      "preferred_identifier": "its own IntAct AC",
      "enumerated_here": false
    },
    {
      "intact_ac": "EBI-2620048",
      "name": "mll_aff4_fusion",
      "description": "Fusion of Human MLL (N -terminus) and Human AFF4 (518-1163) -C terminal region",
      "interaction_count": 6,
      "xrefs": "empty",
      "preferred_identifier": "its own IntAct AC",
      "enumerated_here": true,
      "partners": [
        "MED19 x1 (PMID 21729782, HeLa)",
        "MED26 x4 (PMID 21729782, HeLa)",
        "AFF1 x1 (PMID 20153263, HEK293T)"
      ]
    },
    {
      "intact_ac": "EBI-2620075",
      "name": "mll_aff4_fusion-1",
      "description": "Fusion of human MLL (N-terminal region) and human AFF4 (C terminal region)",
      "interaction_count": 1,
      "xrefs": "empty",
      "preferred_identifier": "its own IntAct AC",
      "enumerated_here": true,
      "partners": [
        "AFF1 x1 (PMID 20153263, HEK293T)"
      ]
    }
  ],
  "total_interaction_records": 47,
  "records_enumerated_in_this_entry": 7,
  "search_stems_tried": [
    "mll_af4",
    "mll_aff1",
    "mll_fusion",
    "mll_enl",
    "mll_af9"
  ],
  "search_stems_returning_records": [
    "mll_enl",
    "mll_af9"
  ],
  "set_is_a_floor": "Only five name stems were tried and two returned records, so four interactors is a lower bound and not an exhaustive enumeration.",
  "why_invisible": "Both interactome gap analyses query by UniProt accession and resolve partner identity with the regex uniprotkb:([A-Za-z0-9._-]+)\\(gene name\\). A fusion construct's MITAB alias field is psi-mi:<name>(display_short)|psi-mi:<AC>(display_long) with NO (gene name) token, so gene_of() returns None and the row is dropped from the union, the read queue and the coverage denominator - silently and without a counter.",
  "leakage_artifact": "PSICQUIC search/query/<accession> is free-text, not id-scoped, so query/Q03164 and query/Q9UHB7 both return the same MED19 interaction EBI-4321349. The read queue's MED19 n_baits of 2 is that double attribution, and neither KMT2A nor AFF4 is what MED19 was tested against.",
  "why_it_matters": "This project's most-quoted self-criticism is that 0 of 24 sampled load-bearing experiments rest on a dedicated human KMT2A::AFF1 experiment, and its standing caveat is that an interactor of KMT2A is not an interactor of the fusion. Forty-seven curated records exist whose participant IS a KMT2A fusion protein, thirty-two of them on MLL-AF9, and this graph has read none of them. They are laboratory chimeras and not endogenous t(4;11) products, so not a substitute for the index fusion - but a strictly better class of evidence than a wild-type KMT2A partner."
}

{
  "measured_2026_09_22": {
    "Q9UHB7_psicquic_rows": 179,
    "Q9UHB7_psicquic_distinct_pmids": 23,
    "Q9UHB7_psicquic_max_pmid": "39251607",
    "Q9UHB7_psicquic_rows_from_40205054": 0,
    "P48426_psicquic_rows": 234,
    "P48426_psicquic_distinct_pmids": 17,
    "P48426_psicquic_max_pmid": "39251607",
    "P48426_psicquic_rows_from_40205054": 0,
    "P48426_rest_indexed_interactions": 320,
    "rest_rows_from_40205054": "present for both accessions"
  },
  "missing_publication": "PMID 40205054, Schaffer LV et al., Nature 2025 - joint biophysical interaction and immunofluorescence map of >5,100 proteins in U2OS, 275 molecular assemblies. A different laboratory, a different cell line and an orthogonal modality from BioPlex.",
  "consequence_for_the_headline_number": "analyses/core-complex-interactome-gap/results.json reported 217 union partners, 89 with an entry (41.0%) and 119 absent; re-run after this tranche it gives 94/217 (43.3%), 112 absent, read queue 20 -> 15, denominator unchanged at 217. Either way it is computed against a mirror whose newest publication is from 2024, so numerator AND denominator are FLOORS and nothing in the results file says so.",
  "which_way_the_error_runs": "The denominator is too small, so unnominated partners exist. The per-partner publications lists are too short, so a row that looks like one platform may have two: CAMKV, PIP4K2A and EPB41L1 each gain a publication between the PSICQUIC snapshot and the REST query. The brief's warning that a 2-paper row may rest on one platform is correct about what the queue can see and understates the evidence for three of this tranche's seven partners.",
  "recommendation": "Move run.py to IntAct REST and CARRY BOTH NUMBERS with a banner, per this project's rule that an endpoint change must be announced as an instrument change rather than a corrected figure. Do not silently re-run.",
  "also_found": "The PSICQUIC tab25 response carries only 15 COLUMNS, so experimental role, biological role and expansion method are unavailable from that endpoint entirely. A filter written against the 25-column MITAB spec drops every row and returns a clean zero - which this beat hit on its first sweep and corrected by printing the column count."
}

{
  "measured_2026_09_22": {
    "findInteractions/EBI-2620048 rows returned": 7,
    "intact interactionCount for EBI-2620048": 6,
    "spurious row": "S100B-NEDD8, interaction AC EBI-26200486, PMID 32814053",
    "second instance": "SPATS1-SARS1 via EBI-26200754 on a query for EBI-2620075"
  },
  "control": "In a spurious row NEITHER participant is the queried interactor. Apply it to every accession-based IntAct query.",
  "trend": "Will worsen: IntAct accessions have grown by an order of magnitude, so short early accessions are prefixes of long recent ones."
}

[
  "Enumerate the four fusion-construct interactors explicitly as extra baits in analyses/core-complex-interactome-gap/run.py - they cannot be discovered by accession, so they must be named.",
  "Make gene_of() record an unresolvable partner in a named counter instead of dropping it silently.",
  "Make every interactome accession filter isoform-tolerant by stripping a trailing -N; measured cost of not doing so is CAMKV's 12 rows seen as 6 and 3 publications seen as 2.",
  "Move to IntAct REST and carry both numbers with a banner rather than correcting 89/217 in place."
]