A73051 has the lowest oncoprotein share of the 25 untreated samples, and its sequencing depth is mid-range (37th-39th percentile), so the reading is not a depth artefact - the coordinator's cycle-135 qualification misidentified enrichment as coverage and was reversed in cycle 138

🔵 **POSITIVELY SUPPORTED 2026-09-24 (cycle 139), not merely un-refuted.** The depth-matched thinning the coordinator demanded has run: **44 of 54 libraries thinned to A73051's in-promoter count shift by a median +0.0003; 39 of 39 hold, 0 collapse; 0 of 54 thinned 95% intervals contain A73051's 1.208.** Combined with A73051 being at the **37th-39th percentile of true sequencing depth**, the depth objection is gone in both its forms. **Your conclusion is now supported by the test that was supposed to break it.** --- 🔵 **KING QUALIFICATION REVERSED 2026-09-24 (cycle 138): your conclusion stands and my qualification of it was wrong.** In cycle 135 I dropped this entry to `confidence: weak`, rewrote its title, and said the 'globally low-oncoprotein' reading was inseparable from library coverage. **The variable I called coverage was the in-promoter enrichment fraction, not sequencing depth.** Against true depth, counted from the deposit's own fragment BEDs: - A73051's KMT2A-N libraries hold **9,101,995 and 9,279,052 fragments - ranks 20 and 21 of 54, the 37th and 39th percentiles** (median 10,300,517). **Your description of them as 'mid-range' was correct.** - The oncoprotein share correlates with true depth at **-0.255** across 54 libraries and **-0.289** across 27 samples - **negative**, where my confound story needed a strong positive. 🟠 **What remains true from the qualification, and it is narrower:** the share does track the **enrichment fraction** at +0.659, and the floor-referenced `onco/neg` ratio tracks it at +0.631, so neither is invariant to how much specific signal a library captured. For a CUT&RUN that is arguably part of the measurement rather than a confound. **Confidence restored to `moderate`**, below the original only because the enrichment relationship is real and the depth-matched down-sampling remains unrun. --- 🔴 **KING QUALIFICATION 2026-09-24 (cycle 135): the measurements here reproduce exactly; the CONCLUSION does not follow from them.** This entry's own control C3 measured **Spearman(oncoprotein share, raw library promoter mean) = +0.659** (p = 6.2e-08, 54 libraries) - above the 0.6 threshold this beat set for itself - and its C8 note then described A73051's libraries as *"mid-range"*. **They rank 7th and 10th lowest of 54, the 13th and 19th percentiles** of a 1.302-18.795 range, which is exactly where a +0.659 correlation predicts a low share. **The floor-referenced `onco/neg` ratio was the escape route and the coordinator tested it: it still correlates with raw coverage at +0.631.** There is no coverage-invariant measure in this deposit. 🔵 **And the confound cannot be subtracted, which is why the verdict is NEITHER rather than a refutation.** In a spike-in-free CUT&RUN, scarce target protein *causes* shallow coverage, so the correlation is equally predicted by this entry's own biological reading. **Abundance and depth are one variable here.** The cheapest discriminator - down-sampling the deep libraries to A73051's coverage - has not been run. See `ana-whether-a73051-is-globally-low-oncoprotein-cannot-be-answered-in-this-deposit`. --- `ana-q147-a73051-is-a-hoxa-negative-irx1-negative-kmt2a-af10-blast` asks in its own open questions: *"Is A73051 globally low-oncoprotein, or HOXA-selectively low?"* **Globally low.** Three independent statistics, in ascending order of how little they depend on the global normalisation: 1. **Canonical class share.** A73051's `Oncoprotein_Target` (827 promoters) share is **1.208x** an average promoter in its own libraries (N1 1.197, N2 1.218) - **rank 1 of the 25 untreated samples**, rank 2 of 27, behind only **152985, the revumenib-exposed aliquot at 1.180x**. The untreated range is 1.208-3.330. 2. **Floor-referenced.** Dividing by the eight-locus non-expressed floor instead of the global mean gives `onco/neg` = **6.96**, third lowest of the 25 untreated - only A80956 (3.28, depth-INADEQUATE) and TB920488 (5.85) are lower - against an untreated maximum of 201.9 in TB995291 and an infinite value in TB000694 whose floor reads exactly zero; the drug-exposed 152985 is 1.65. A second denominator, same direction. 3. **N-over-C.** The deposit's own criterion for fusion-specific occupancy is N/C > 1.75. A73051's `Oncoprotein_Target` N/C is **1.048** - **rank 1 of the 25 untreated**, rank 2 of 27 behind 152985 (0.965). **16 of 27 samples exceed 1.75.** A73051 shows essentially no N-terminal excess over the C-terminal antibody anywhere on the canonical programme. And against a coverability-matched null, A73051's canonical class is **significantly DEPLETED** - 0.667x its matched null, two-sided permutation p = 0.001 in both N libraries.