Across every same-condition H2AK119ub replicate pair in GSE260742 - nine pairs, three arms, both pairing shapes - rank concordance over the shared bin universe runs 0.6481 to 0.8796 and never reaches 0.88. The floors this thread scores against are all computed from pairs that are only moderately concordant.
also called nothing here is a tight replicate · Spearman 0.65 to 0.88 · the floors rest on moderate pairs
Repeat measurements in this dataset rank the genome's regions only moderately alike - better than chance, well short of a close match. Every reliability yardstick this project uses is built from such pairs.
The same thing, for a scientist
Spearman rank correlation of within-sample shares over each pair's shared bin universe (1.11-1.17 M bins), computed by one function for all pairs: | pair | arm | shape | Spearman | |---|---|---|---:| | C1_DMSO | PCGF1-KO | marker | 0.8308 | | C1_SGC | PCGF1-KO | marker | 0.7742 | | C1_VTP | PCGF1-KO | marker | **0.6481** | | NT_DMSO | non-targeting | marker | 0.7614 | | NT_SGC | non-targeting | marker | **0.8796** | | NT_VTP | non-targeting | marker | 0.6913 | | C-IDENT DMSO/SGC/VTP | cross-title | marker | 0.8436 / 0.8412 / 0.8259 | **The range is 0.65-0.88 and the distribution does not separate by arm.** The widest floor (MLLAF9 Control, 37.04% at D1) and the tightest (PCGF1-KO, 1.89%) come from pairs whose rank concordance overlaps, so **the floor is not simply tracking concordance** - two pairs can rank the genome equally alike and still redistribute their share mass across strata very differently. **Why that is the interesting part.** It means the arm-to-arm floor span is not explained away as 'one arm's data is noisier'. Rank agreement is similar; what differs is where the share mass sits relative to the external stratifier's top bins.
The calibration this supplies, which the thread did not have
Before this beat, no number in the thread said how well a GSE260742 'replicate' pair agrees. The C-IDENT control needed exactly that to interpret its 0.83, and the only reason it could return a verdict is that these nine values were computed alongside it. Any future claim in this deposit that two tracks are or are not the same library should be read against 0.65-0.88.
It is also a warning about the opposite inference: 0.83 between two tracks here does not mean they are near-duplicates.
THE RENORMALISATION BOUND, AND IT APPLIES TO EVERY NUMBER IN THIS ENTRY. Within-sample shares sum to 1 and GSE260742 has NO USABLE SPIKE-IN (Q117 verified zero hits for the entire spike-in vocabulary across all 419,824 characters of its 158 sample records, and no scale factor is deposited), so a purely local change leaks into every other stratum by arithmetic. A FLOOR COMPUTED THIS WAY CANNOT DISTINGUISH 'the bulk moved' from 'the peaks moved and the shares renormalised'. A TIGHTER NUMBER IS NOT THEREBY A CLEANER ONE.
| statistic | Spearman of within-sample shares over the shared 2kb-bin universe |
|---|---|
| floor tracks concordance | false |
spearman by pair — the numbers behind the chart
{
"C1_DMSO": 0.8308,
"C1_SGC": 0.7742,
"C1_VTP": 0.6481,
"NT_DMSO": 0.7614,
"NT_SGC": 0.8796,
"NT_VTP": 0.6913,
"C_IDENT_DMSO": 0.8436,
"C_IDENT_SGC": 0.8412,
"C_IDENT_VTP": 0.8259
}range
[ 0.6481, 0.8796 ]
- What would IDR over called peaks say about these same pairs? The deposit ships TDF only, so peak calling needs the BAMs, which are deposited for only some samples.
- Does the pair with the LOWEST concordance (C1_VTP, 0.6481) also give the tightest floor (0.40% at D1)? It does - which is the sharpest single piece of evidence that the floor and the concordance are measuring different things, and it deserves its own analysis.
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset
Spearman rank correlation of within-sample 2 kb-bin shares over the shared universe, for the six GSE260742 NTvKO same-condition H2AK119ub replicate pairs, is 0.6481, 0.6913, 0.7614, 0.7742, 0.8308 and 0.8796; the three cross-title C-IDENT comparisons give 0.8259, 0.8412 and 0.8436. No pair reaches 0.88 except NT_SGC at 0.8796.
What this evidence cannot show
Spearman over ~1.15 M bins of which the great majority carry low coverage is dominated by the low-coverage bulk and is NOT a peak-level concordance measure; ENCODE would use IDR over called peaks, which this deposit's TDF-only form does not permit. These values calibrate each other and should not be compared to published ChIP-seq concordance figures computed differently. SYSTEM: murine MLL-AF9 (KMT2A::MLLT3) myeloid blood cells, Mus musculus, mm10, one lab, one deposit - NOT the infant B-lineage KMT2A::AFF1 disease this graph is about. n = 3 conditions per arm, one cell system; the spread across conditions is reported so stability is visible and IS NOT A CONFIDENCE INTERVAL. THE RENORMALISATION BOUND, AND IT APPLIES TO EVERY NUMBER IN THIS ENTRY. Within-sample shares sum to 1 and GSE260742 has NO USABLE SPIKE-IN (Q117 verified zero hits for the entire spike-in vocabulary across all 419,824 characters of its 158 sample records, and no scale factor is deposited), so a purely local change leaks into every other stratum by arithmetic. A FLOOR COMPUTED THIS WAY CANNOT DISTINGUISH 'the bulk moved' from 'the peaks moved and the shares renormalised'. A TIGHTER NUMBER IS NOT THEREBY A CLEANER ONE.
Evidence that argues against this (2)
supplies the calibration range without which C-IDENT's 0.83 could not be read
shows the arm-to-arm floor span is not explained by one arm being noisier
the concordance range that makes the span puzzling in the first place