Oncofusion

start / graph / absence-q125-gse260742-has-no-bcor-knockout-so-half-of-prediction-5s-disjunction-has-never-been-tested

Analysisstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q125-the-h2ak119ub-arm-of-the-pcgf1-knockout

hyp-q117's prediction 5 says 'in a PCGF1 OR BCOR knockout' and GSE260742's only genetic perturbation is the PCGF1 knockout: across all 158 sample records the strings BCOR, BCORL1, KDM2B, RING1 and RNF2 appear zero times as a perturbation, so a verdict on PCGF1 is not a verdict on the prediction as written

also called the untested half of the disjunction · no BCOR knockout in the series

The prediction named two genes that could be removed. Only one of them was, in the only dataset that could test it. So half the prediction is answered and half has never been looked at.

The same thing, for a scientist

**A disjunctive prediction is not settled by testing one disjunct, and this corpus has just cashed one.** `hyp-q117`'s prediction 5 reads *'in a PCGF1 **or BCOR** knockout...'*. This cycle scored it for PCGF1 [E1]. **Bounded in the archive, not in the literature.** Over all 419,824 characters of GSE260742's 158 sample records, re-fetched this session: the only genotype terms that appear are `PCGF1 KO`, `NTvKO`, `NT`, `CLONE1` and `C1`. **No sample is a BCOR, BCORL1, KDM2B, RING1A/RING1B/RNF2, PCGF2 or PCGF4 knockout**, and the series' only guide pair in the paper's key-resources `Oligos` sheet for the mouse arm is a single PCGF1 guide with no second guide and no rescue construct [E1]. **Why BCOR is not interchangeable with PCGF1, which is what makes the untested half matter.** Q117 measured both proteins at the fusion's eviction quartiles in this same series: at the most-evicted quartile **BCOR rises 1.4433-fold and RING1B 1.3808-fold, while across Q2-Q4 BCOR stays at 1.2701-1.3304 and RING1B falls to 0.8070-0.8792** [E2]. **BCOR's in-fill is large everywhere and RING1B's is specific to the evicted quartile** - so the two subunits are not behaving as one module, and a BCOR knockout has a different predicted phenotype from the PCGF1 knockout this cycle measured. **And BCOR loss is the clinically relevant disjunct**, being recurrent in myeloid leukaemia, whereas PCGF1 loss is a laboratory resistance genotype.

Go deeper
What would close it
  • A deposited BCOR- or BCORL1-knockout arm with a menin inhibitor and an H2AK119ub or fusion-occupancy readout. No cross-series GEO search for one was run in this beat and that is the cheapest next step.
  • The same design in a BCOR-mutant primary or patient-derived sample. BCOR loss is recurrent, so the genotype exists in patients and does not need engineering.
  • Failing both, the prediction should be re-written as two predictions so that a PCGF1 verdict does not read as a verdict on BCOR. That is a wording change the coordinator can make with the cashed text this cycle proposes.
Facts and figures
absence classnever-measured
absence class noteFor GSE260742 this is an absence of the DEPOSIT and it is a never-measured absence of the experiment: no BCOR-knockout sample exists in any of the 158 records and no BCOR measurement of any kind is deposited for the NTvKO cultures, so closing it needs new cell engineering or a BCOR-mutant primary sample - wet work, not compute. NOT bounded against the literature: no PubMed or cross-series GEO search for a BCOR knockout with a menin inhibitor and a chromatin readout was run in this beat, and that search is the cheapest thing that could downgrade this to measured-not-searched.
absence class assigned2026-09-23 (Q125)
searchedall 158 !Sample_title, !Sample_characteristics_ch1 and !Sample_supplementary_file_ fields of GSE260742 (419,824 characters, re-fetched 2026-09-23)
terms
[
  "BCOR",
  "BCORL1",
  "KDM2B",
  "RING1",
  "RNF2",
  "PCGF2",
  "PCGF4",
  "knockout",
  "KO",
  "sgRNA",
  "CRISPR"
]
what would close it
[
  "a deposited BCOR or BCORL1 knockout with a menin inhibitor and an H2AK119ub or fusion-occupancy readout",
  "the same design in a BCOR-mutant primary sample",
  "splitting prediction 5 into two predictions so a PCGF1 verdict is not read as a BCOR verdict"
]
How this could bite us back
  • BCOR loss is recurrent in myeloid leukaemia and PCGF1 loss is not. This cycle tested the laboratory genotype and left the clinical one untested, which is the direction this corpus's evidence usually runs and is worth naming rather than absorbing.
Still open
  1. Does any series anywhere deposit a BCOR or BCORL1 knockout with a menin inhibitor? No cross-series GEO search was run and it is one query.
  2. Would a BCOR knockout give a LARGER effect than PCGF1, given that Q117's BCOR in-fill is broad while RING1B's is eviction-specific? The two subunits' different spatial patterns make this a real prediction rather than a formality.
Every source behind this entry (2)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    Across all 158 sample records of GSE260742 (419,824 characters, re-fetched 2026-09-23 and byte-identical to Q124's copy), the only genetic perturbation represented is the PCGF1 knockout clone C1 against a non-targeting-sgRNA control NT; no BCOR, BCORL1, KDM2B, RING1A, RING1B/RNF2, PCGF2 or PCGF4 knockout sample exists, and the paper's key-resources Oligos sheet lists a single mouse PCGF1 guide with no second guide and no rescue.

    What this evidence cannot show

    This bounds the DEPOSIT and not the literature: no PubMed or cross-series GEO search for a BCOR-knockout menin-inhibitor chromatin experiment was run in this beat, so the claim is 'not in this series', not 'nowhere'. The guide list is quoted from Q124's reading of 41556_2025_1859_MOESM6_ESM.xlsx and was not independently re-read here. Sample-record text is the only route: `antibod` occurs zero times in the same 419,824 characters, so these records are sparse and an unlabelled perturbation could in principle be present.

    Evidence that argues against this (1)
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    Q117 measured, in this same series and murine culture, that at the most-evicted quartile of the fusion's top-0.1% set BCOR's VTP/DMSO within-sample share ratio is 1.4433 and RING1B's 1.3808, while across the other three quartiles BCOR stays at 1.2701-1.3304 and RING1B falls to 0.8070-0.8792 - so BCOR's in-fill is broad and RING1B's is eviction-specific.

    What this evidence cannot show

    Read from a committed results.json, not re-measured. Q117's quartile analysis is recorded there as its own DEVIATION 3. These are occupancy ChIPs of proteins, n = 1 per condition, with no spike-in.

    Evidence that argues against this (1)
Connected to