Analysis: does the fusion's target set sit at lamina-associated domain borders? A free overlay of published LAD maps on published fusion occupancy
also called LAD border overlay · HOXA lamina test · NEAT1 subtype expression query
LAD borders are marked by CG islands — the same feature the fusion reads. This checks, using only data already published, whether the fusion's landing sites sit at the edges of the silenced blocks that touch the nuclear envelope, and whether the HOXA gene cluster is one of them.
The same thing, for a scientist
Two cheap public-data questions bundled into one runnable analysis, both currently unanswered and both cheap enough that not running them is indefensible. **Question 1 (LAD borders).** LADs are bordered by CTCF sites, outward-facing promoters and CpG islands; the fusion's CXXC domain reads non-methylated CpG islands. The analysis intersects published Lamin B1 DamID LAD calls — the human fibroblast reference map and the closest available haematopoietic map, from Lamin B1 DamID in Jurkat T cells — with published KMT2A::AFF1 occupancy, testing whether fusion peaks are enriched within a defined window (e.g. ±50 kb) of LAD borders relative to an accessibility- and CpG-matched null, and separately whether the HOXA cluster falls inside a LAD, at a border, or in open inter-LAD space in each available cell type. **Question 2 (NEAT1 by subtype).** In the same run, query NEAT1 expression in paediatric ALL by molecular subtype in TARGET and St. Jude Cloud counts, testing whether KMT2A-rearranged cases differ from other subtypes — the single cheapest test of the paraspeckle hypothesis. Both questions are bundled because both are public-data overlays on the same cohorts and neither justifies a separate analysis pipeline. The deliverable is a short report with an honest negative if that is the answer.
Why bundle two unrelated questions
Because both are of the same kind: a compartment hypothesis in this beat with no direct evidence, whose first-order test is an overlay on data that already exists. Keeping them in one analysis node makes the cost explicit — a few days of one person's time for two compartments — and prevents each from being deferred individually on the grounds that it is 'not the main question'.
The cell-type caveat applies to both and should be stated in the report rather than buried. LAD maps are strongly cell-type specific and no fetal-liver HSC or LMPP LAD map exists, so a fibroblast reference map can only support a weak, hypothesis-generating conclusion; the Jurkat map is haematopoietic but is a mature T-lineage line. Neither is the cell of origin. A negative result therefore constrains less than a positive one would, and the report should say by how much.
Pre-specified interpretation
LAD borders. If fusion peaks are enriched at LAD borders beyond CpG-matched expectation, the lamina entry moves from 'no evidence' to 'a genomic association worth an experiment', and a Lamin B1 DamID in a degron line becomes justified. If not, localization-nuclear-lamina-lads-kmt2a-r can be closed as a compartment and the beat stops spending attention on it.
HOXA position. If HOXA sits in or at the edge of a LAD in a progenitor-like cell, then the fusion's central target is a locus whose normal state is peripheral silencing, and the fusion's action can be described as forcing a lamina-proximal locus into the active interior — a specific, imageable claim.
NEAT1. If NEAT1 is significantly lower in KMT2A-rearranged infant ALL than in other subtypes, that matches the PML::RARA precedent in direction and justifies NEAT1 RNA-FISH on archival blasts. If it is unchanged or higher, the paraspeckle-as-differentiation-brake analogy is not supported and should be retired from the graph rather than left as an open question indefinitely.
| script | analyses/lad-border-overlap-and-neat1-subtype/run.py |
|---|---|
| status | planned |
| result summary | not yet run; outputs (a) enrichment of fusion peaks within +/-50 kb of LAD borders against a CpG- and accessibility-matched null, (b) the LAD status of the HOXA cluster in each available cell type, (c) NEAT1 expression by paediatric ALL subtype with KMT2A-rearranged versus others, effect size and CI |
| figure | /figures/lad-border-and-neat1-subtype.svg |
| cost | days of one analyst's time; no new data generation |
inputs
[ "published Lamin B1 DamID LAD calls (human fibroblast reference; Jurkat T-cell activation series), lifted to hg38", "published KMT2A::AFF1 CUT&RUN/ChIP-seq occupancy (SEM, RS4;11), hg38", "hg38 CpG islands, CTCF sites, matched accessibility for the null model", "HOXA cluster coordinates", "TARGET and St. Jude Cloud paediatric ALL RNA-seq counts with molecular subtype labels" ]
closes entries if negative
[ "localization-nuclear-lamina-lads-kmt2a-r", "localization-paraspeckles-neat1-kmt2a-r" ]
- A negative here is weaker than it looks because no LAD map exists for the cell of origin; reporting it as 'the lamina is not involved' would overstate what the data can support, and the report must carry that qualification prominently.
- Bundling two questions risks one being run carelessly to get to the other; the report should state the analyst-time split so that a thin NEAT1 query cannot be passed off as having settled the paraspeckle question.
- Is a fibroblast LAD map informative at all for a fetal haematopoietic progenitor? If the two available maps disagree about HOXA, the honest output is that the question needs a haematopoietic LAD map before it can be answered, and that is a useful result.
- Should the NEAT1 query be extended to paired diagnosis-relapse samples? If NEAT1 moves at relapse, it becomes a candidate marker of the differentiation state rather than of the fusion.
- Would the same overlay run against menin-inhibitor-treated occupancy data show fusion peaks retreating from, or toward, LAD borders? That converts a static overlay into a dynamic one at no extra cost if treated occupancy data exist.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Human lamina-associated domains are gene-poor, transcriptionally repressive domains of 0.1-10 Mb sharply bordered by CTCF binding sites, outward-facing promoters or CpG islands, mapped genome-wide by DamID.
Domain organization of human chromosomes revealed by mapping of nuclear lamina interactions · Guelen L, Pagie L, Brasset E, Meuleman W, Faza MB, Talhout W, Eussen BH, de Klein A, Wessels L, de Laat W, van Steensel B · Nature 2008What this evidence cannot show
Human lung fibroblasts; LAD maps are cell-type specific and this reference map is not haematopoietic, which bounds the strength of any conclusion drawn from it.
Evidence that argues against this (1)
- This is a methodological objection to the analysis rather than to the LAD map, and it is the one that matters for the proposed overlap statistic. Kind et al. used a molecular contact-memory approach to follow LADs in single cells and found that in any one nucleus only about 30 per cent of LADs are actually at the periphery; contacts with the lamina are intermittent; and after mitosis LAD positioning is not detectably inherited but is stochastically reshuffled. A population DamID LAD map is therefore a per-locus probability, not a per-cell geometry, and 'distance from a LAD border' computed against it describes no individual nucleus. An overlap between KMT2A target genes and population LAD borders can be entirely real and still correspond to no physical arrangement in any blast; the analysis needs a permutation null built from single-cell contact frequencies, not from the consensus map. Read: abstract only. Kind J et al. 2013
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Lamin B1 DamID with FISH in Jurkat T cells shows constrained release of lamina-associated enhancers and genes from the nuclear envelope during T-cell activation, with TADs repositioning as units, providing the closest available haematopoietic LAD map.
Constrained release of lamina-associated enhancers and genes from the nuclear envelope during T-cell activation facilitates their association in chromosome compartments · Robson MI, de Las Heras JI, Czapiewski R, Sivakumar A, Kerr ARW, Schirmer EC · Genome Research 2017What this evidence cannot show
A mature T-cell leukemia line under activation, not a myeloid or B-lymphoid progenitor and not a KMT2A-rearranged context.
Evidence that argues against this (1)
- A population DamID map is an ensemble average that no single nucleus realises, so an overlap test against 'LAD calls' is testing against a construct. Kind et al. tracked genome-lamina contacts in single cells with a molecular-contact-memory approach and found that only about 30% of LADs sit at the nuclear periphery in any one nucleus; individual LADs make intermittent molecular contact with the lamina while remaining constrained to the periphery; and LAD positioning is not detectably inherited through mitosis but is stochastically reshuffled. Contact was linked to repression and H3K9me2 in single cells, with G9a as a regulator. A 'within +/-50 kb of a LAD border' window therefore has a different meaning cell by cell, and an enrichment computed against the ensemble map can be produced or destroyed by the fraction of cells in which the locus is peripheral at all - which this analysis as designed cannot measure. Kind J et al. 2013
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
NEAT1 is repressed in de novo APL and by PML-RARA, rises during ATRA-induced differentiation and is required for it, making a subtype-stratified NEAT1 expression query in paediatric ALL the direct analogue test in KMT2A-rearranged disease.
Inhibition of long non-coding RNA NEAT1 impairs myeloid differentiation in acute promyelocytic leukemia cells · Zeng C, Xu Y, Xu L, Yu X, Cheng J, Yang L, Chen S, Li Y · BMC Cancer 2014What this evidence cannot show
APL and a single cell line; the transfer to KMT2A-rearranged disease is an analogy whose direction is not even certain, which is precisely why the cheap query should be run rather than argued about.
Evidence that argues against this (1)
- The NEAT1/APL strand of this analysis inherits a problem from the paraspeckle literature. Nakagawa et al.'s NEAT1 knockout mice are viable and fertile with no phenotype beyond the loss of paraspeckles, and paraspeckles themselves form only in a subpopulation of cells in response to particular triggers. A NEAT1 expression change in APL or AML is therefore not evidence about a nuclear compartment, and should not be used to anchor a compartment-level analysis. Read: abstract only. Nakagawa S et al. 2011
- ← related toThe nuclear lamina and lamina-associated domains - evidence grade for KMT2A: weak and indirect, not none
the counted, dated negative (0 records, 2026-09-19) is what makes the reanalysis rather than the experiment the next step
- related to →The nuclear lamina and lamina-associated domains - evidence grade for KMT2A: weak and indirect, not none
Counter-evidence (Kind 2013, PMID 23523135): only ~30% of LADs sit at the periphery in any one nucleus and positioning is stochastically reshuffled at mitosis, so a consensus DamID map is a per-locus probability and the overlap statistic needs a single-cell null
public-data overlay that closes or opens the lamina question
subtype-stratified NEAT1 query is the cheapest test of the paraspeckle analogy
- depends on →Dataset: TARGET via the GDC - open expression and mutations, but the fusion calls themselves are controlled
uses the TARGET paediatric ALL expression cohort
single-cell contact memory: ~30% of LADs peripheral per nucleus and positioning reshuffled at mitosis (Kind 2013), so a border-overlap test scores against an ensemble construct