ExperimentstrongQ114-suv39h1-at-the-locus

The deposit this graph filed as 'four SUV39H1 ChIP-seqs' is actually 77 samples in MCF-7 carrying the ENTIRE H3K9 axis in one cell type and one pipeline - writer SUV39H1, writer G9a/EHMT2, reader HP1-alpha/CBX5 (wild-type AND two phospho-mutants), co-repressor KAP1/TRIM28, erasers KDM4B and KDM4C, the mark H3K9me3, and an shKDM4B+shKDM4C eraser-knockdown arm

also called GSE139196 full sample census · the H3K9 axis in MCF-7 · SUV39H1 deposit is bigger than its entry says

A dataset this project had catalogued as four experiments on one enzyme turns out to contain seventy-seven, covering every protein that writes, reads, removes or is the silencing tag - all in the same cells, on the same day, with the same method. That is the comparison this project has never been able to make, because until now every piece came from a different cell type.

The same thing, for a scientist

**What the deposit actually contains.** GSE139196 (`ChIRC-seq Reveals Promoter Antisense RNAs and HP1alpha as Key Regulators of Promoter Pause Release...[ChIP-Seq]`, the ChIP-Seq sub-series of SuperSeries GSE139199) holds **77 samples**, all MCF-7, all with a matched minus-/plus-estradiol (100 nM, 40 min) contrast, all through one HOMER pipeline to hg38 [E1]: | antigen as labelled in the deposit | n | |---|---:| | H3K9me3 | **8** | | POLR2A (RPB1) | 8 | | NELFA | 5 | | HP1A (CBX5) | **4** - two of them Bio-ChIP on CBX5 **mutant 4A** and **mutant 4E** | | G9a (EHMT2) | **4** | | KAP1 (TRIM28) | **4** (+2 more labelled bare `KAP1`) | | KDM4B | **4** | | KDM4C | **4** | | SUV39H1 (KMT1A) | **4** | | H3K27me3 / H3K36me3 / H3K4me3 / H3K56ac / H3K9K14ac | 4 each | | HEXIM1 | 4 | | POLR2A Ser2P | 4 | | H3K122ac | 2 | **Why this matters more than the SUV39H1 samples on their own.** This graph's H3K9 argument is assembled from different cells: the mark in RPE (Q108, GSE210480), H3K9me1 in K562, SETDB1 and CBX5 in K562/GM12878/H1/HepG2 (Q110, ENCODE), EHMT2 in K562/A549/HepG2, TRIM28 in K562/HEK293, and now SUV39H1 in DIPG and MCF-7. **Every cross-protein comparison the graph makes is therefore also a cross-cell-type comparison**, and `ana-q110-h3k9-writer-and-reader-occupancy-at-the-kmt2a-locus` says so in its own open question. GSE139196 removes that confound for six of the eight players at once - writer, second writer, reader, co-repressor, two erasers and the mark - in one cell, one pipeline, one condition pair [E1]. **And it contains a genetic perturbation arm nobody here has noticed.** Four H3K9me3 samples are `shctrl` and two are `shKDM4B_shKDM4C` (GSM4133820-23), i.e. **H3K9me3 measured with both KDM4 erasers knocked down against a matched shControl**. This graph runs a live, unresolved dispute about whether H3K9 demethylase activity at the *KMT2A* locus is protective or permissive (`mechanism-kdm3b-loss-shifts-the-kmt2a-anchor-from-h3k9me2-to-h3k9me1` versus `mark-h3k9me3`), and the only perturbation data it has ever had for that axis is the KDM3B/G9a siRNA series in RPE. **The limit, stated first and not in a caveat: MCF-7 is an oestrogen-receptor-positive breast adenocarcinoma.** Nothing in this deposit is haematopoietic, none of it is *KMT2A*-rearranged, and the estradiol contrast is an oestrogen-signalling perturbation with no counterpart in infant B-ALL. What the deposit buys is **internal consistency across the H3K9 axis**, not relevance to the index disease. **Relationship to existing entries.** `exp-suv39h1-chip-seq-in-mcf7-with-and-without-estradiol` describes the four SUV39H1 samples and is correct about them; it does not record that the other 73 exist. This entry is the census, not a replacement, and it also corrects that entry's `genome_build` field (see below).

Go deeper

A correction to exp-suv39h1-chip-seq-in-mcf7-with-and-without-estradiol, with the source

That entry's data.genome_build reads: "NOT STATED in the per-sample data_processing lines; must be verified from the file's own chrom table before any coordinate is used." The first clause is false. Every GSE139196 sample's !Sample_data_processing block carries two independent statements of the build 1:

Read alignment: Bowtie2 (Langmead et al., Nature Methods 2012) was employed to align
  ChIP-seq and ChIRC-seq samples to hg38 genome ...
Genome_build: hg38

Read from acc.cgi?acc=GSE139196&targ=gsm&form=text&view=brief on 2026-09-22, present in all four SUV39H1 samples and in the sixty-odd others inspected. My own dispatch brief repeated the entry's claim as a premise, which is how a wrong field in one entry becomes a wrong premise in the next beat's brief.

The second clause is right and this beat obeyed it anyway. A bedGraph has no chrom table, so the build was established independently from each file's own maximum coordinate per chromosome, in both directions (chr1, chr2, chr19 and chrX are longer in hg19; chr11 is longer in GRCh38), before any coordinate was read. The metadata and the file agree. The rule survives the correction: what changes is only that the deposit is not silent.

What could be run from this deposit without any new wet work

All of it is deposited as per-sample bedGraph on hg38 and all of it is re-readable with the Q114 pipeline, which converts a bedGraph to a bigWig and reads it with the same code path as a CUT&RUN track.

Writer-reader-eraser coincidence at one locus in one cell. SUV39H1, EHMT2, CBX5, TRIM28, KDM4B, KDM4C and H3K9me3 over the same intervals, same null, same day.
The eraser-knockdown arm. H3K9me3 shKDM4B+shKDM4C versus shCTRL, at the KMT2A exon-11 anchor and genome-wide - a direct test of whether KDM4 activity holds the mark down at this element, in the only system where the experiment exists.
The HP1-alpha phospho-mutant arm. Bio-ChIP on CBX5 4A and 4E. gene-cbx5 asks whether HP1-alpha's function is chromodomain-dependent; these are N-terminal phospho-site mutants rather than chromodomain mutants, so they do not answer that question - but they are the only published HP1-alpha mutant occupancy maps in a human cell and nobody here knows they exist.
A 40-minute ligand time point. The only condition-contrasted SUV39H1 occupancy data in existence.

Facts and figures

design	census of an existing public deposit, not a new experiment: all 77 per-sample SOFT records of GSE139196 parsed and tabulated by labelled antigen and condition
readout	which proteins of the H3K9 axis have deposited genome-wide occupancy in ONE human cell type with a matched perturbation
controls	no antibody clone, no input track and no IgG identified in the sub-series by this beat; the minus-estradiol and shCTRL arms are the internal contrasts
cost tier	low
timeline	already deposited; every arm is compute-only with the Q114 bedGraph-to-bigWig pipeline
accession	GSE139196 (ChIP-Seq sub-series of SuperSeries GSE139199)
n samples	77
perturbation arm	GSM4133820, GSM4133821 (shCTRL) versus GSM4133822, GSM4133823 (shKDM4B + shKDM4C), H3K9me3, minus estradiol
hp1a mutant arm	GSM4133790 (CBX5 mutant 4A) and GSM4133791 (CBX5 mutant 4E), Bio-ChIP, minus estradiol
genome build	hg38, STATED TWICE in every per-sample data_processing block (Bowtie2 ... to hg38 genome; Genome_build: hg38) - and independently re-derived from each file's own maximum coordinate per chromosome, because a bedGraph has no chrom table
cell type	MCF-7, ER-positive luminal breast adenocarcinoma
cell state	hormone-deprived (5% charcoal-stripped FBS, >=3 days) versus 100 nM 17-beta-estradiol for 40 minutes; ethanol vehicle controls
lineage	breast epithelium - NOT haematopoietic, NOT KMT2A-rearranged
shown for fusion	NO - nobody has looked in any KMT2A fusion system, and this is a breast line
corrects	exp-suv39h1-chip-seq-in-mcf7-with-and-without-estradiol, data.genome_build, which says the build is NOT STATED in the per-sample processing lines

antigen counts

antigen counts — the numbers behind the chart

{
  "H3K9me3": 8,
  "POLR2A (RPB1)": 8,
  "NELFA": 5,
  "HP1A (CBX5)": 4,
  "G9a (EHMT2)": 4,
  "H3K27me3": 4,
  "H3K36me3": 4,
  "H3K4me3": 4,
  "H3K56ac": 4,
  "H3K9K14ac": 4,
  "HEXIM1": 4,
  "KAP1 (TRIM28)": 4,
  "KDM4B": 4,
  "KDM4C": 4,
  "POLR2A (RPB1) Ser2P": 4,
  "SUV39H1 (KMT1A)": 4,
  "H3K122ac": 2,
  "KAP1": 2
}

suv39h1 samples

[
  "GSM4133857",
  "GSM4133858",
  "GSM4133859",
  "GSM4133860"
]

h3k9me3 condition pair used by q114

[
  "GSM4585206 (-E2)",
  "GSM4585207 (+E2 40 min)"
]

eraser knockdown arm answered cycle 74

{
  "question": "does knocking down both H3K9me3 erasers raise the mark at the KMT2A locus?",
  "answer": "NO",
  "positive_control_fires": "the arm works and reproduces the depositors' own figure (PMID 34194047 ED Fig. 7p): H3K9me3 at the active-promoter +1 nucleosome moves 21.73 -> 57.43 percentile, delta +35.71, while the 1 kb promoter window moves -4.79 - confined to the +1 nucleosome exactly where the paper puts it",
  "at_the_locus": "exon-11 reads 0.00 in all four tracks; the 90 kb gene body goes shCTL 7.60/7.70 -> shKD 2.56/2.64, delta -5.05, with within-arm spreads of 0.10 and 0.08",
  "interpretation": "the methyl-poor gene body is NOT an erasure product - it gets relatively poorer when the erasers go. Because the peak caller depth-normalises, the honest wording is that the locus does not join in a genome-wide rise rather than that it falls absolutely",
  "resolved_by": "the cycle-68 beat; verified and recorded by the king 2026-09-23"
}

How this could bite us back

A deposit with no antibody clone, no input and no knockout control for any of its 77 arms is a deposit in which every negative could be an antibody failure and every positive could be cross-reactivity. Any intervention argument built on the KDM4-knockdown arm inherits that, and the registry row on antibody clones being the identifier applies in full.
The estradiol contrast makes this deposit attractive as a 'perturbation' dataset, but oestrogen signalling has no counterpart in infant B-ALL. Reading a 40-minute E2 redistribution as a model for any leukaemia perturbation would be the cross-system generalisation this project documents as its commonest synthesis failure.

Still open

Does H3K9me3 rise at the KMT2A exon-11 anchor when both KDM4 erasers are knocked down? GSM4133822/23 against GSM4133820/21 answers it with compute alone, and it is the only eraser-perturbation arm in any archive that touches this mark at this element. Not run by this beat, which spent its budget on the writer.
Do SUV39H1, EHMT2, CBX5 and TRIM28 occupy the same sites in MCF-7? The whole axis is in one cell here for the first time, and the pairwise overlap has never been reported - including by the paper the SuperSeries is attached to, which is about pause release.
Are the two samples labelled bare 'KAP1' the same antibody as the four labelled 'KAP1 (TRIM28)'? The deposit gives no clone for any target, so this cannot be settled from GEO and would need the corresponding author.

Every source behind this entry (2)

E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset
GSE139196 contains 77 MCF-7 samples covering SUV39H1, EHMT2/G9a, CBX5/HP1A (including 4A and 4E mutants), TRIM28/KAP1, KDM4B, KDM4C and H3K9me3 with matched +/- estradiol contrasts, and every sample's data_processing block states Genome_build: hg38.
GSE139196: ChIRC-seq Reveals Promoter Antisense RNAs and HP1alpha as Key Regulators of Promoter Pause Release and of Enhancer : Architectural RNA Interactions [ChIP-Seq] - per-sample SOFT records · Yang F, et al. · NCBI GEO 2019
What this evidence cannot show
Counts are of SAMPLES as labelled by the submitter in !Sample_characteristics chip antibody, parsed from the full 77-sample SOFT dump fetched 2026-09-22. Two records are labelled bare 'KAP1' and four 'KAP1 (TRIM28)'; they are counted separately above because the deposit does not state that they are the same antibody. No antibody clone is given for any target, which is a real limit on every one of these arms.
Evidence that argues against this (1)
- A sample label is the submitter's claim about what was immunoprecipitated. With no clone, no input and no knockout control identified in the sub-series, none of these 77 labels is independently verifiable, and a mislabelled or cross-reactive antibody would look exactly like a biological result. source
E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
GSE139196 carries no PMID of its own; the SuperSeries GSE139199 carries PMID 34194047 and names GSE139196 as one of four sub-series.
Shape of promoter antisense RNAs regulates ligand-induced transcription activation. · Yang F, et al. · Nature 2021
PMID 34194047 DOI
What this evidence cannot show
Verified 2026-09-22 by fetching both series records with targ=self&form=text&view=brief: GSE139196 has no !Series_pubmed_id line, GSE139199 has !Series_pubmed_id = 34194047 and !Series_relation = SuperSeries of: GSE139196. PubMed metadata for the PMID was fetched live and the DOI read from PubmedData/ArticleIdList only. The abstract was read; the full text was not.
Evidence that argues against this (1)
- The paper is about promoter antisense RNAs and HP1-alpha in pause release. There is no evidence that it analysed the SUV39H1, KDM4B, KDM4C or EHMT2 arms at all, so attributing conclusions about those arms to it would be an over-attribution - the deposit is being used here as data, not as a published result. source

Connected to

refutes →GSE139196: SUV39H1 ChIP-Seq in MCF-7 with and without 40 minutes of estradiol - a ligand-inducible SUV39H1 occupancy experiment, filed under the retired symbol KMT1A in its own sample titles
corrects data.genome_build: hg38 IS stated per sample, twice
tests →H3K9me3 - measured in KMT2A::AFF1 cells, in a supplementary figure, and it rises GENERALLY where H3K27ac falls SPECIFICALLY in the same experiment
8 H3K9me3 samples incl. an shKDM4B+shKDM4C eraser-knockdown arm, same cells as the writer
tests →CBX5 / HP1-alpha (P45973, 191 aa) - the H3K9me3 reader that recruits the writer, the writer's destroyer and the second writer; and the one member of that circuit with a published pro-leukaemic role, as part of an MLL leukaemia stem-cell programme that immortalises myeloid progenitors WITHOUT Hoxa or Meis
the only published HP1-alpha mutant (4A, 4E) occupancy maps in a human cell
tests →EHMT2 / G9a - the H3K9me1/2 writer that sets whether the KMT2A LOCUS ITSELF rearranges, and whose loss delays MLL-AF9 AML in a supplementary figure nobody can find
G9a occupancy in MCF-7, same cells and pipeline as SUV39H1 and H3K9me3
tests →TRIM28 / KAP1 / TIF1beta (Q13263) - the KRAB-ZFP corepressor scaffold in the pan-MLL-fusion core, with the weakest index-fusion p-value but a leukaemic role in another fusion family
KAP1/TRIM28 occupancy in the same deposit
tests →KDM4C (JMJD2C/GASC1) — recruited by the retained KMT2A N-terminus, drugged in a KMT2A::AFF1 line, and invisible to CRISPR
KDM4C occupancy and a double KDM4B/KDM4C knockdown H3K9me3 arm
tests →SUV39H1 (O43463, KMT1A) - the H3K9me3 writer: a menin effector whose binding region OVERLAPS the drug pocket at its N-terminus, a tumour suppressor in MLL-AF9 AML, recruited to double-strand breaks by CTCF, and never localised at the KMT2A locus in any cell type
the four SUV39H1 (KMT1A) samples, the only condition-contrasted SUV39H1 occupancy data in existence
supports →RUN: ten ENCODE peak sets and fourteen signal tracks read at the KMT2A locus in two blood cells - not one called peak for SETDB1, CBX5, CBX1, CBX3 or TRIM28 anywhere in the gene, while G9a/EHMT2 and TRIM28 both sit on the KMT2A PROMOTER and neither touches the breakpoint cluster region
removes the cross-cell-type confound that entry names in its own open question, for six of the eight players
← testsThe H3K9me3 this graph argues about at the KMT2A exon-11 anchor is not deposited by SUV39H1: in SU-DIPG13 the mark sits at the 57.8th percentile of class-matched CTCF anchors while the writer reads an EXPLICIT ZERO in the same nuclei, and genome-wide the two barely track each other (Spearman 0.148, 0.078)
its four unanalysed EHMT2 samples test prediction 3 with compute alone
← depends onThe MCF-7 arm, and the clause that saved it: SUV39H1 at the exon-11 anchor reaches the 96.3rd percentile of 56,171 same-cell-type CTCF anchors in one replicate - but all six non-anchor KMT2A intervals are elevated too (medians 73-87), so the pre-registered FIRES clause does not fire, the verdict is NEITHER, and four replicates of one antibody in one cell line span 51.8 to 96.3 at the same 147 bp
the deposit census that identifies these four samples and their conditions
← testsRUN: G9a/EHMT2 reads the 95.8th percentile of 56,171 same-cell-type CTCF anchors at the KMT2A promoter in all four MCF-7 ChIP-seq tracks, and a median 33.42nd at the exon-11 anchor - so the cheapest test of hyp-q114's third prediction REFUTES it
the G9a arm of that experiment entry, run
← testsRUN: KAP1/TRIM28 has a called peak overlapping the KMT2A promoter in MCF-7 in both hormone conditions, 267 bp of it inside the K562 TRIM28 peak Q110 found - and its nearest chr11 peak to the exon-11 CTCF summit is 52,230 bp away
the KAP1 arm of that experiment entry, at peak level, plus the correction that the deposit does contain two peak files
← related toANSWERED from the paper's own methods, not from the corresponding author: the `dsgf` token that splits GSE139196's four SUV39H1 tracks into the two high and the two low ones is DSG-plus-formaldehyde double crosslinking - so the 44.5-percentile spread Q114 reported is a protocol variable, not replicate noise
a protocol variable encoded in that deposit's filenames, unread until now, that partitions one antigen's tracks
← related toKDM4B (JMJD2B) - the weakest of the four H3K9me3 demethylases, one of the two erasers knocked down in the only KDM4-perturbation arm in any archive that touches this mark
four KDM4B ChIP-seq tracks and one of the two shRNAs of that deposit's eraser-knockdown arm
← testsRUN, with its positive control: knocking down KDM4B and KDM4C together raises H3K9me3 at the +1 nucleosome of 2,000 active promoters by 35.7 percentile points, moves the KMT2A promoter's +1 nucleosome by the same 35.3, and moves the 90 kb KMT2A gene body 5.05 points the WRONG way - so the methyl-poor domain is not an erasure product
the eraser-knockdown arm of that experiment entry, run, with the genome-wide positive control Q114 said it lacked