AFF1 and fusion occupancy transfer between infants at Spearman 0.82-0.92, but H3K4me3 transfers better at 0.93-0.97 - so the transferable part is mostly promoter structure, not fusion targeting
Where the fusion sits ranks similarly from one child's leukaemia to another's. But a general marker of switched-on genes ranks even more similarly, so most of that agreement is just the fact that active genes are active everywhere.
The same thing, for a scientist
**This project's standing pattern is that between-system ORDERING transfers while MAGNITUDE does not. GSE252378 is the first chance to test it inside one deposit, one protocol, one genome build and one laboratory. The result is that the ordering half needs a benchmark it has never been given.** **Pairwise Spearman over 19,404 within-sample promoter shares** [E1]: | factor | range across all available pairs | |---|---| | IgG (the floor) | **0.121 - 0.175** | | DOT1L | 0.488 - 0.613 | | ENL | 0.709 - 0.767 | | **AFF1** | **0.820 - 0.893** (includes SEM vs each patient: 0.820-0.868) | | KMT2A N-terminal | 0.834 - 0.920 | | **H3K4me3** | **0.932 - 0.970** | **The IgG floor is 0.12-0.18, so the correlations are not an accessibility artefact** - the pre-registered B3 'neither' branch does not fire on the IgG test. **But H3K4me3, which by the paper's own use is the negative control that does not distinguish fusion from wild-type sites, transfers better than any fusion-related track.** So an AFF1 or KMT2A-N correlation of 0.85 between two samples is not evidence that fusion targeting transfers; it is below what a generic active-promoter mark achieves in the same samples. **The right reading, and it is narrower than the project's standing claim.** Between-sample ordering agreement is real and high, and its ceiling is set by promoter structure rather than by the fusion. A transfer claim about fusion occupancy needs the H3K4me3 value beside it, or it is quoting the promoter set. **A cell line is not an outlier here.** SEM's AFF1 profile correlates with the three patients at 0.820 / 0.851 / 0.868, inside the range the patients achieve with each other (0.831-0.893). On this readout the line-versus-primary gap is not visible at all - which is worth saying next to the HOXA result, where it is the whole story. **Genome-wide ordering and individual loci give opposite answers about how good a model SEM is, and both are correct.**
Why a rank correlation cannot see the thing that matters
Spearman over all 19,404 promoters is dominated by the several thousand promoters that are active in every haematopoietic cell. The fusion-target set is 971 promoters, 5% of the total, so a complete reshuffling of the fusion targets moves the global rho by very little. That is exactly what happens between p279 and p318: global AFF1 rho 0.831 - high - while the fusion-target Jaccard between the same two samples is 0.117. The two statistics are not in conflict; they answer different questions, and only the second one is about the fusion.
| script | analyses/q142-aff1-in-primary-infants/analyze.py |
|---|---|
| status | run |
| result summary | Ordering transfers (AFF1 0.82-0.89) but below the H3K4me3 benchmark (0.93-0.97) with an IgG floor of 0.12-0.18. |
| pre registered outcome | B1 magnitude criterion not met; B3 not fired on IgG but the H3K4me3 ceiling qualifies B1/B2 |
- What is the between-sample rank correlation of a factor that genuinely does not transfer, in this same assay? Nothing in the deposit supplies that lower benchmark except IgG, which is too sparse to be a fair one.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset
Between samples in GSE252378, promoter-share Spearman is 0.121-0.175 for IgG, 0.820-0.893 for AFF1, 0.834-0.920 for KMT2A N-terminal and 0.932-0.970 for H3K4me3.
GSE252378 - AutoCUT&RUN and AutoCUT&Tag profiles of 36 MLL-rearranged leukemia samples (171 GSMs, hg38) · Janssens DH, Henikoff S (Fred Hutchinson Cancer Center / Van Andel Institute) · NCBI Gene Expression Omnibus 2024What this evidence cannot show
Two patients, three states, one cell line. Promoter windows only. All values are within-sample shares; no spike-in. Antibody-matched pairs were used for every comparison, but the AFF1 comparisons average two antibodies that disagree with each other at rho 0.71-0.78 within a sample, so the AFF1 row carries reagent variance the H3K4me3 row does not.
Evidence that argues against this (1)
- H3K4me3 is a far more abundant target than any of these factors, so its higher reproducibility could be signal-to-noise rather than biological universality; benchmarking a low-abundance factor against a high-abundance mark may be unfair and would understate how well fusion occupancy transfers. Janssens DH et al. 2024
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The source paper reports that genomic enrichment of the oncoprotein is highly variable between samples and subject to dynamic regulation.
MLL oncoprotein levels influence leukemia lineage identities · Janssens DH, Duran M, Otto DJ, Wu W, Xu Y, Kirkey D, Mullighan CG, Yi JS, Meshinchi S, Sarthy JF, Ahmad K, Henikoff S · Nature Communications 15:9341 2024What this evidence cannot show
The paper's variability statement is measured with its own oncoprotein score over its own peak set, which is a different instrument from a promoter-share rank correlation.
Evidence that argues against this (1)
- High global ordering agreement and high oncoprotein-score variability are both true and are measured on different scales; a reader could reasonably say this entry's benchmark comparison does not engage the paper's variability claim at all. Janssens DH et al. 2024
source deposit
SEM sits inside the patient-to-patient range on global ordering
global rho 0.83 and fusion-target Jaccard 0.12 in the same pair
the other half of the same pre-registered question
those figures are single-channel; on the fusion-specific ratio the question is unresolvable