Oncofusion

start / graph / ana-rbp-bioid2-proximity-map-k562

Analysisstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q20-proximity-interactome

The 50-RBP BioID2 proximity map in K562 - the first modern proximity dataset this graph can reach, and AFF1 is in it as prey

also called RBP BioID2 map · IM-30059 · PXD041608 · Khoroshkin 2024 proximity map

A 2024 study attached a biotin-tagging enzyme to 50 RNA-handling proteins inside human leukaemia cells and recorded everything each one brushed past. AFF1 - half of the fusion this project studies - shows up in that record, and nobody here had ever looked.

The same thing, for a scientist

Khoroshkin et al. fused BioID2 to **50 human RBPs in K562 cells**, captured each neighbourhood by streptavidin pulldown and mass spectrometry in biological triplicate against unlabelled controls, and deposited the curated bona fide pair list to IMEx as **IM-30059** (proteomics PXD041608; sequencing GSE225809, confirmed by `elink dbfrom=pubmed&db=gds`) [E1]. IntAct serves **19,403 proximity rows** from that deposition, resolving to **50 distinct baits and 2,682 distinct preys**, median 5 baits per prey, mean 7.23, maximum 40 [E2]. Every row is PSI-MI **MI:1314** (proximity-dependent biotin identification), **MI:1060 spoke-expanded**, host organism annotated `taxid:9606(human-k_562)`, MI-score 0.27 [E3]. **No super elongation complex component, no KMT2A-complex component and no fusion protein is a bait.** AFF1, AFF4, ELL, MLLT1, CDK9, CCNT1, KMT2A, MEN1 and DOT1L appear only in the prey column. That single fact frames everything else this beat found.

Go deeper
What the experiment actually is

BioID2 was fused to each of 50 RBPs and expressed as a transgene in K562, a chronic-myeloid-leukaemia line carrying BCR::ABL1 and no KMT2A rearrangement. Biotinylated proteins were recovered on streptavidin and identified by mass spectrometry, three biological replicates per bait against unlabelled control designs 1.

Three consequences the rest of this beat inherits:

  1. Wild-type proteins only. There is no fusion protein anywhere in this dataset. Anything it says about AFF1 is about endogenous full-length AFF1.
  2. Overexpressed, tagged baits, one isoform each. The authors state this limitation themselves 1.
  3. Bait-centric, and the baits are all RBPs. A protein can only appear if some bait's neighbourhood reached it, so the map samples the nucleus from the RNA side. Nothing in it was chosen because it mattered to transcription elongation.
The deposition is the usable object, not the paper

The strings AFF1, AFF4, ELL and KMT2A do not appear in the paper's body text at all - they are in supplementary data files and in the IMEx deposition only 1. That is the mechanism behind this beat's central negative result about searching (see ana-proximity-weak-spot-readjudication).

The route that works is the method code: detmethod:"MI:1314" intersected with an accession. The route that fails is any query built from gene symbols and method names against titles and abstracts 4.

Control: is the method filter real?

Per the project's standing rule that a new instrument must pass a known-positive before its negatives are reported: IntAct serves 64,945 MI:1314 rows overall, and 259 for LMNA (P02545), a canonical BioID subject. The filter returns large numbers where large numbers should exist, so the small numbers reported across this beat (AFF1 4, MEN1 3, MLLT3 0) are measurements rather than a broken query 3.

Facts and figures
scriptnone (database query)
statusrun
pmid39251607
imexIM-30059
pridePXD041608
n rows19403
n baits50
n bait experiments52
n preys2682
median baits per prey5
mean baits per prey7.23
max baits per prey40
cell lineK562 (BCR::ABL1 chronic myeloid leukaemia; no KMT2A rearrangement)
methodBioID2, PSI-MI MI:1314, spoke-expanded MI:1060
result summaryA 50-bait human proximity map exists in a leukaemia cell line and contains AFF1, AFF4, ELL, MLLT1, CDK9, CCNT1, KMT2A, MEN1 and DOT1L - all as prey. No SEC or KMT2A-complex protein has ever been a proximity-labelling bait.
titleSystematic identification of post-transcriptional regulatory modules.
authorsKhoroshkin M, Buyan A, Dodel M, et al.
venueNat Commun
year2024
title provenanceTitle/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading.
inputs
[
  "IntAct PSICQUIC (EMBL-EBI), retrieved 2026-09-19",
  "NCBI E-utilities elink pubmed->gds",
  "Europe PMC fullTextXML PMC11385195"
]
geo
[
  "GSE225809",
  "GSE225808",
  "GSE225807",
  "GSE225806"
]
How this could bite us back
  • This dataset is a wild-type, non-KMT2A-rearranged, overexpression map in one cell line. Anything the graph takes from it is an import across cell line, across fusion status and across construct, and should carry the confidence of an import.
  • Reading a proximity hit as complex membership is the specific error this beat exists to avoid. Every entry in this folder states the distinction; if the entries are ever summarised, the distinction is the first thing lost.
  • A drug programme that nominated a target from this map would be nominating an abundance-weighted nuclear neighbour, not a binding partner.
Still open
  1. The authors assign each of the 50 BAITS to a regulatory module. AFF1 is a prey and therefore has no module. Does re-running their integrated distance on the prey side place AFF1 with the splicing baits or with the transcription baits?
  2. Are the raw unfiltered hit lists in PXD041608 richer for AFF1 than the curated IMEx subset, and does AFF1 appear in more than four bait neighbourhoods before the bona fide filter?
  3. Is anything like this map available in a KMT2A-rearranged line? K562 has no KMT2A rearrangement, so the entire dataset is a wild-type control for this project's disease.
  4. Does the Human Cell Map (192 BioID baits, HEK293) contain AFF1 as a prey? Its data sit behind a JavaScript application with no REST endpoint found on 2026-09-19.
Every source behind this entry (4)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Khoroshkin et al. fused BioID2 to 50 human RNA-binding proteins in K562 cells and captured each protein's neighbourhood by streptavidin pulldown and mass spectrometry in biological triplicate against unlabelled controls; the curated list of bona fide BioID protein pairs was deposited to IMEx as IM-30059, the proteomics to PRIDE as PXD041608 and the sequencing to GEO as GSE225809. The authors state that RBP overexpression, BioID2 fusion and the use of a single isoform per protein may bias the data.

    PMID 39251607DOI
    What this evidence cannot show

    Full text read from PMC11385195 (body element present, 141,821 characters), not abstract only. AFF1 and the other elongation factors appear nowhere in the body text; they are reachable only through the supplementary data and the IMEx deposition. K562 is a BCR::ABL1 chronic myeloid leukaemia line with no KMT2A rearrangement, so nothing here is measured in the index disease.

    Evidence that argues against this (2)
    • A direct head-to-head comparison of BioID against an affinity-purification protocol optimised for chromatin-associated complexes found the two approaches identified FEW SHARED partners and enriched for distinct biological processes. A proximity hit is therefore not a weaker version of a co-purification hit; it is a different measurement, and the overlap between the two is small by construction. Lambert JP et al. 2015
    • The founding BioID report defines the method as identifying proximal AND interacting proteins and demonstrates labelling of proteins that are near the bait without binding it. Promiscuous labelling of the bait's vicinity is the designed behaviour of the reagent, not an artefact to be argued away. Roux KJ et al. 2012
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    IntAct serves 19,403 proximity rows from this deposition, resolving to 50 distinct baits and 2,682 distinct preys, with a median of 5 baits per prey, a mean of 7.23 and a maximum of 40; 43.7% of preys are recovered by four or fewer baits.

    What this evidence cannot show

    Retrieved 2026-09-19. Counts are of curated rows, not of the raw mass-spectrometry hit lists, which are larger; the IMEx record is the authors' filtered bona fide set. Two baits (TAF15, ZC3H11A) carry a second experiment identifier, giving 52 bait-experiment combinations for 50 proteins.

    Evidence that argues against this (2)
    • The largest published BioID survey of a human cell (192 baits, HEK293) shows that proximity recovery is organised by subcellular compartment: a prey's bait profile is dominated by where it lives rather than by what it binds. A nuclear elongation factor recovered by nuclear RNA-processing baits is the expected result of shared compartment occupancy alone. Go CD et al. 2021
    • The distribution of baits-per-prey limits every specificity claim drawn from this dataset. With a median of 5 and 43.7% of preys at four or fewer, a protein recovered by four baits sits near the middle of the distribution and is not a specific hit. The dataset's own shape argues against reading small recovery counts as selectivity. IntAct molecular interaction database (EMBL-EBI) 2026
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    Every row of this deposition is PSI-MI MI:1314 (proximity-dependent biotin identification), spoke-expanded under MI:1060, with the bait carrying experimental role MI:0496 and the partner MI:0498, host organism annotated taxid:9606(human-k_562), and an IntAct MI-score of 0.27. As a positive control on the method filter, IntAct serves 64,945 MI:1314 rows in total and 259 for LMNA (P02545).

    What this evidence cannot show

    MI-score 0.27 is what a single-method, single-publication proximity record scores by construction; it carries no information about the strength of any individual pair. Spoke expansion means the deposited interactions are bait-to-prey assignments, not observed binary contacts.

    Evidence that argues against this (1)
    • IntAct is one database with one curation policy. The Human Cell Map (192 BioID baits in HEK293) is not served through this route, so a zero from this query bounds IntAct's holdings rather than the world's proximity data. Go CD et al. 2021
  4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    The GEO depositions of this study are reachable by elink from the PubMed identifier: elink dbfrom=pubmed&db=gds&id=39251607 returns uids 200274273, 200274252, 200274049, 200225809, 200225808, 200225807 and 200225806, of which 200225809 is GSE225809, the accession named in the paper's own data-availability statement.

    What this evidence cannot show

    A procedural check, not a biological result. Recorded because the project's GEO route was corrected one cycle ago and this is an independent confirmation that the elink entry point works where the esearch one does not.

    Evidence that argues against this (1)
    • Agreement between elink and the data-availability statement is not general. The project's own measurement found elink alone resolves 37.5% of depositions and the statement is needed for the rest, so neither route bounds the other. IntAct molecular interaction database (EMBL-EBI) 2026
Connected to