The menin interface cannot be what decides where KMT2A::AFF1 sits: within one patient the fusion's target promoters were almost entirely replaced while the menin-binding motif was unchanged

🟠 **STATUS CHANGED AGAIN 2026-09-24 (cycle 131), and the honest position is weaker than either previous banner.** The cycle-128 note below removed this entry's evidence on the strength of a retention figure that does not exist. q146 restores the **ordering** but shows the effect is **1.33× over a stratified floor, at the 37th percentile of unrelated pairs, present in every unrelated AFF1 pair, and institution-confounded.** 🔴 **So the addressing claim has no clean quantitative support from either direction.** q144's invariant-interface/variable-occupancy result is about **magnitude** and still stands; this entry's claim is about **address** and should be read as **open**. --- 🟠 **RE-GROUNDED 2026-09-24 (cycle 128). THE CONCLUSION STANDS; ITS ORIGINAL EVIDENCE DOES NOT.** This entry argued from target-set **replacement** within one patient — the Jaccard 0.117 now superseded, since q144 measures the same switch keeping **1,894 of 2,045 calls**. 🔵 **q144 reaches the same conclusion independently and more strongly, from a different quantity:** the menin-binding N-terminus is **intact and identical in all 27 patients across six fusion partners**, occupancy varies by **two orders of magnitude**, and **occupancy does not sort by partner**. *The interface is a permission, not a dial.* 🔴 **Stated precisely, because the two are not the same claim:** q144's result is about **magnitude**, this entry's is about **address**. The addressing claim currently has **no surviving quantitative support in this corpus** and should be read as open rather than established. See `ana-two-beats-measured-the-same-lineage-switch-and-got-opposite-answers-and-the-better-powered-one-wins`. --- **What this beat can and cannot say about the menin interface, stated as a bound rather than as a result, because no menin was measured** [E3]. **The constraint.** The menin-binding motifs MBM1/MBM2 sit in the retained KMT2A N-terminus and are identical in `p279` and `p318` - the same patient, the same fusion allele, the same interface. Between those two states the N/C-defined fusion-target promoter set changes almost completely: **Jaccard 0.117**, lower than the 0.149 between two unrelated infant B-ALLs [E1]. EBF1 goes from N/C 3.09 to 1.02; HOXA9/HOXA10 goes from KMT2A-N 3.16x mean promoter to 0.15x; PAX5 and MEIS1 do not move [E1]. **Therefore an unchanged interface is compatible with a nearly completely changed target set.** The menin interface may be **necessary** for fusion recruitment - nothing here tests that - but it cannot be the thing that **selects** which promoters the fusion occupies. The selector has to be something that differs between the two states: chromatin accessibility, the lineage transcription-factor landscape, CpG/H3K36me2 address availability, or fusion abundance. **This is a constraint on a model, not evidence against a drug.** A menin inhibitor works by removing a necessary contact; it does not need the contact to be the selector. The practical consequence is narrower and sharper: **the set of promoters a menin inhibitor would evict the fusion from is a property of the patient's current lineage state, not of the fusion.** A pharmacodynamic gene list built in a B-ALL is the wrong list for the AML that the same patient's disease becomes - and for this patient, that transition happened. **A second, smaller observation.** The MEN1 promoter itself is not a fusion target in any of the three states, and its N/C ratio is below 1 in all three (0.67 / 0.78 / 0.51) - i.e. what sits on MEN1's own promoter is wild-type KMT2A, not the fusion [E1]. There is no fusion-driven MEN1 autoregulatory loop visible in primary infant material. **What would refute the menin-interface hypothesis outright, scoreably** [E2]: - **Menin CUT&RUN in a KMT2A::AFF1 infant finding menin at fewer than half of that sample's N/C-defined fusion-target promoters.** The fusion would then be sitting at most of its sites without menin, and the interface would not be a general requirement. The denominator for that test is now defined and is roughly 1,000 promoters per sample. - **Menin occupancy that is as flat across fusion and wild-type sites as AFF1's is here (ratio 0.56-1.45 against ENL's 1.90-2.78).** Menin would then carry no more fusion information than the partner protein does, and every menin-occupancy argument in this corpus would inherit the same defect this beat found in AFF1 occupancy. - **A menin inhibitor evicting the fusion from wild-type KMT2A promoters as much as from fusion-target promoters**, which would make the interface a general KMT2A-chromatin requirement rather than a fusion-specific one. **None of these can be run in this deposit**, because it contains no menin library in any KMT2A::AFF1 sample; the only menin tracks are in one KMT2A::ENL patient under drug.