Scored at complex level, coherence predicts nothing (rho -0.016 over 719 complexes) - but the Fanconi core complex is 2nd of 719 and the HBO1/MOZ acetyltransferase modules are 1st

ana-fa-module-coherence reported rho = -0.764 within the Reactome Fanconi set and a citric-acid-cycle second place at -0.657, and explained both by observing that the signal sits in a sub-complex - the FA core complex, and seven SDH subunits inside TCA. Its standing conclusion was that the pathway is the wrong unit.

The script states its criterion before running: "a rho near zero, or SDH and the FA core sitting mid-pack - either of which would say the sub-complex reading was a story told about two pathways after the fact."

The between-complex Spearman is -0.016 across 719 complexes. That criterion fires. Tightly co-dependent complexes are not more KMT2A-selective than loosely co-dependent ones, and the coherence-drives-selectivity reading has no support at the unit it was supposed to work best at.

The cleanest single refutation is the electron transport chain.

These are the most coherent sets in the entire sweep and none is in the top 30. Coherence and selectivity come apart most sharply exactly where coherence is highest, which is the opposite of what the mechanism predicts.

The Fanconi anaemia ubiquitin ligase complex is position 2 of 719 - FAAP100, FAAP20, FANCA, FANCB, FANCC, FANCE, FANCF, FANCG, FANCL - with a median selectivity rank of 125 against a uniform expectation of about 8,864, and a best member at rank 24. Nine members, so this is not a small-set artefact. The finding is real and it is extreme; only the explanation was wrong.

SDH comes out much smaller than last cycle made it look. It is 38th of 719, which is genuinely selective - the median complex sits at 8,175 - but complexes I, III and IV are also more selective than the median complex. The whole electron transport chain trends selective, and SDH is the strongest of four rather than a distinct phenomenon. The previous cycle's reading, that seven of the twelve most selective TCA members are SDH subunits, is still true; what changes is that it stops being evidence for a sub-complex mechanism and becomes one more instance of a broad mitochondrial trend.

It also had to be found by hand. The script's own lookup for SDH returned nothing, because Complex Portal files it as "Mitochondrial respiratory chain complex II" and the search was by the name succinate dehydrogenase. A name-keyed lookup for a complex fails the same way a symbol-keyed literature search does, and the script reported "NEITHER is present" while the complex was sitting at position 38.

Eight of the top twelve are HBO1, MOZ or MORF complexes. KAT7 (HBO1) and KAT6A (MOZ) are both already in this graph as genes, reached by other routes; an unbiased sweep of 719 complexes putting their modules at rank 1 is independent corroboration that nothing here was fitted to.

The honest discount is large, and it is visible in the table. Those eight complexes are variants sharing subunits - the repeated median rank of 173 is the tell - and the whole top twelve spans only 52 distinct genes, of which 15 are Fanconi and 26 are mediator. The acetyltransferase block is roughly eleven genes shuffled into eight rows: KAT7, KAT6A, KAT6B, ING4, ING5, JADE1/2/3, MEAF6, BRPF1/3, BRD1. Eight of twelve is one observation, not eight.

Size is a real confound and it is bounded rather than dismissed. Complexes of 4-5 members are 410 of 719 (57%) but 35 of the top 50 (70%) - over-represented. They are also 17 of the bottom 50, which is the signature of a noisy median rather than of a selective biology. So the n=4 rows deserve suspicion. Core mediator at n=26 and position 11, and the FA complex at n=9 and position 2, do not, and they are the two results here that are safe to build on.