Oncofusion

start / graph / exp-suv39h1-chip-seq-in-mcf7-with-and-without-estradiol

Experimentestablished confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.Q113-bounded-absence-audit

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

also called GSE139196 · SUV39H1 ChIP MCF-7 · ChIRC-seq SUV39H1 arm · GSE139196 - build corrected 2026-09-22

In 2021 a group measured where the H3K9 silencing enzyme sits on DNA in breast cancer cells, before and forty minutes after a hormone signal. Because the enzyme is the same one this project argues about, the data are usable here even though the cells are not blood cells.

The same thing, for a scientist

Four SUV39H1 ChIP-Seq samples (GSM4133857-60) in MCF-7, two untreated and two at 40 min estradiol, deposited in the ChIP-Seq sub-series GSE139196 of the SuperSeries GSE139199 and published in *Nature* 2021 (PMID 34194047), a study of promoter antisense RNAs and HP1 in promoter pause release. ChIP-Atlas indexes the same four as SRX7030945-48 under antigen SUV39H1. Supplementary files are per-sample bedGraphs. Two things make this worth an entry of its own rather than a footnote. First, it is a **condition-contrasted** SUV39H1 occupancy experiment - plus and minus a ligand at a 40-minute timescale - which is the only published SUV39H1 dataset in which the enzyme's distribution is measured in two states of the same cell. Second, its sample titles read `ChIP on SUV39H1 (KMT1A)`, and **KMT1A is a spelling this project's SUV39H1 literature searches never tried**; the alias is harmless here because the GEO title index matched on SUV39H1 anyway, but it is the exact shape of the MRE11/MRE11A and NBN/NBS1 traps already in the registry.

Go deeper
What is actually contrasted

Two biological conditions, 40 minutes apart, same line, same antibody label, two samples each 1. Forty minutes is short enough that a difference in SUV39H1 occupancy is redistribution rather than a change in protein level, which is the property that makes the pair informative: it separates where SUV39H1 can be from where SUV39H1 goes when a signal arrives. No other deposited human SUV39H1 experiment has a second condition.

The dimensions the PI asks for

Cell type: MCF-7, ER-positive luminal breast adenocarcinoma - not haematopoietic, and the furthest of the two available SUV39H1 systems from this project's biology. Cell state: hormone-deprived versus 40 min oestrogen-stimulated; the state variable is a nuclear-receptor signal, not a differentiation state. Subcellular and subnuclear localization: chromatin-bound only; the parent paper's subject is promoter-proximal pausing and enhancer-architectural RNA contacts, so its own subnuclear framing is promoter and enhancer compartments rather than heterochromatin. Epigenetics: SUV39H1 occupancy alone in this sub-series; the SuperSeries GSE139199 carries the matched ChIRC-seq and PRO-seq arms, so a co-analysis with nascent transcription is possible without leaving the deposit 2.

The spelling, recorded because it is the third proven failure mode

KMT1A is SUV39H1's systematic methyltransferase name and appears in these very sample titles. KMT1A[tiab] returns 22 PubMed records; KMT1A[tiab] AND KMT2A[tiab] and KMT1A[tiab] AND MLL[tiab] both return zero with non-empty translations, so in this instance the untried alias would not have changed any conclusion 3. Recording it anyway, because the failure mode is that nobody checks: a census normalised to the current HGNC symbol reads a well-measured protein as never-measured, and this protein's own deposited data are labelled with the other name.

Facts and figures
designSUV39H1 ChIP-Seq in MCF-7, two samples without estradiol and two at 40 min estradiol
readoutgenome-wide chromatin occupancy; per-sample bedGraph supplementary files
controlsno input or IgG track identified in the sub-series by this beat; the minus-estradiol pair is the internal contrast
cost tierlow
timelinealready run and deposited; re-analysis is compute only
accessionGSE139196 (ChIP-Seq sub-series of SuperSeries GSE139199)
pmid34194047
genome buildhg38. CORRECTED 2026-09-22 by Q114. This field previously read: "NOT STATED in the per-sample data_processing lines; must be verified from the file's own chrom table before any coordinate is read" - and the first clause was FALSE. Every GSE139196 sample's !Sample_data_processing block states the build TWICE: 'Read alignment: Bowtie2 ... to hg38 genome' and 'Genome_build: hg38', read from acc.cgi?acc=GSE139196&targ=gsm&form=text&view=brief on 2026-09-22. The second clause was right and was obeyed anyway: a bedGraph has no chrom table, so Q114 established the build independently from each file's own maximum coordinate per chromosome, in both directions (chr1/chr2/chr19/chrX are longer in hg19, chr11 is longer in GRCh38), before any coordinate was read. The metadata and the files agree. This wrong field became a premise in the Q114 dispatch brief, which is how an error in one entry propagates into the next beat's instructions.
antibody clonenot given in the deposit - only the target label 'SUV39H1 (KMT1A)'
cell typeMCF-7, ER-positive luminal breast adenocarcinoma
cell statehormone-deprived versus 40 min oestrogen-stimulated
localizationchromatin-bound fraction only
subnuclear localizationnot measured; the parent study's framing is promoter-proximal and enhancer compartments rather than heterochromatin
epigenetic readoutSUV39H1 occupancy; matched ChIRC-seq and PRO-seq arms exist in the SuperSeries
shown for fusionNO - nobody has looked in any KMT2A fusion system, and this is a breast line
alias notethe sample titles use KMT1A, SUV39H1's systematic name, which no SUV39H1 literature query in this corpus had tried before 2026-09-22
samples
[
  "GSM4133857 (-E2)",
  "GSM4133858 (-E2)",
  "GSM4133859 (+E2 40 min)",
  "GSM4133860 (+E2 40 min)"
]
chip atlas srx
[
  "SRX7030945",
  "SRX7030946",
  "SRX7030947",
  "SRX7030948"
]
How this could bite us back
  • Using a breast-cancer oestrogen-response dataset as the evidence base for a leukaemia heterochromatin claim is exactly the partner-imported-evidence pattern this project keeps finding in its own corpus. Any conclusion drawn here must name MCF-7 in the claim, not only in the caveat.
Still open
  1. Does SUV39H1 occupancy redistribute at 40 minutes of estradiol, and if so is the redistribution at promoters or in heterochromatin? The deposit contains both conditions and the comparison appears not to have been published as a SUV39H1 result in its own right. ANSWERED IN CYCLE 66, AND THE ANSWER IS ABOUT THE DEPOSIT: no redistribution is measurable, because the replicate noise exceeds any condition effect. Over 56,171 MCF-7 CTCF summit windows (ENCFF942TCG) the ACROSS-condition Spearman is 0.358 while the within-condition replicate correlations are 0.326 (-E2) and 0.207 (+E2) - two replicates of one condition resemble each other LESS than the two conditions resemble each other. At the KMT2A exon-11 anchor the four samples read the 51.78th, 69.05th, 90.81st and 96.29th percentile, with the conditions interleaved. Note that the two samples carrying the unexplained 'dsgf' token (GSM4133857, GSM4133859) are the two HIGH tracks and the two without it are the two LOW ones, a perfect split - so the spread may be a protocol effect rather than replicate noise. See ana-q114-the-four-mcf7-suv39h1-tracks-barely-agree-with-each-other-and-estradiol-changes-nothing.
  2. Is the deposited bedGraph hg19 or hg38? The deposit does not say, and every coordinate claim from it is blocked until that is read from the file. ANSWERED AND THE PREMISE WAS WRONG, CYCLE 66. The deposit DOES say: 'Genome_build: hg38' plus 'Bowtie2 ... to hg38 genome' in every sample's data_processing block. Independently verified from each file's own maximum coordinate per chromosome by Q114 (verdict GRCh38, hg19 excluded, 25 chromosomes in the bedGraph, 49,840,662 lines in GSM4133857). See exp-q114-gse139196-holds-the-whole-h3k9-axis-in-one-cell-type.
  3. Both available human SUV39H1 occupancy systems are non-haematopoietic. Is a CD34+ or leukaemic SUV39H1 CUT&RUN feasible with the CST #8729 clone used in the DIPG deposit, and would that single experiment close the whole SUV39H1 arm of this graph's H3K9 argument?
  4. PARTLY ANSWERED IN CYCLE 66 AND THE REMAINING OPEN PART NAMED: Q114 read all four SUV39H1 samples at the KMT2A locus and the whole 90 kb gene body is elevated uniformly (81.7th-86.8th percentile of same-cell CTCF anchors in all four tracks) while H3K9me3 in the same cells is DEPLETED over the same 90 kb (15.8th and 10.3rd percentile) - so the deposit's SUV39H1 arm does not localise to heterochromatin or to promoters at this locus. It also fails the heterochromatin positive control (0.93-1.15x at ZNF274 3') that the same deposit's H3K9me3 tracks pass at 3.79-5.41x. See ana-q114-in-mcf7-the-whole-kmt2a-locus-is-elevated-and-the-anchor-is-not-picked-out and ana-q114-in-mcf7-h3k9me3-is-depleted-across-the-kmt2a-locus-while-the-writer-looks-elevated.
Every source behind this entry (3)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset

    Four SUV39H1 ChIP-Seq samples in MCF-7, two without and two with 40 minutes of estradiol, are deposited in GEO with the ChIP antibody named in the sample characteristics.

    What this evidence cannot show

    Read from the SOFT records on 2026-09-22. The characteristics line names the target but not a catalogue number or clone, so rule 74's reagent identifier is NOT available for this deposit - which is a real limitation relative to GSE225028, where it is.

    Evidence that argues against this (1)
    • The per-sample data_processing lines do not state a genome build, unlike GSE225028. Any coordinate read off these bedGraphs must have its build verified from the chrom table per file, which the registry already records as a live hazard because one deposit can mix builds and pad states across its own files. NCBI Gene Expression Omnibus 2021
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The SUV39H1 ChIP arm sits inside a SuperSeries whose other arms are ChIRC-seq and PRO-seq in the same system, so nascent-transcription co-analysis is available without a new deposit.

    PMID 34194047
    What this evidence cannot show

    The SuperSeries composition is read from the GEO series records, not from the paper. Which of its 101 samples are PRO-seq versus ChIRC-seq was not enumerated by this beat.

    Evidence that argues against this (1)
    • The paper's subject is HP1 and promoter antisense RNA at ligand-responsive promoters, so the SUV39H1 arm was collected to serve that argument. A track collected to test a promoter hypothesis may be under-powered in constitutive heterochromatin, which is where a KMT2A-locus question would look. NCBI Gene Expression Omnibus 2021
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    KMT1A, the alias these samples are titled with, retrieves 22 PubMed records and none of them pairs SUV39H1 with KMT2A or MLL.

    What this evidence cannot show

    Run through scripts/pubmed_safe.py on 2026-09-22; both zeros carried a non-empty querytranslation and warninglist 'No items found.'. Titles and abstracts only.

    Evidence that argues against this (1)
    • A zero on an alias that returns only 22 records overall is a weak test - the alias is rare enough that the conjunction was always going to be small. The informative version of this check is on aliases that are the MAJORITY token, which is what the registry's 'the approved symbol is the minority token' row is about, and KMT1A is not one of those. HUGO Gene Nomenclature Committee 2026
Connected to