Analysis (planned): is CLEC12A a selective dependency in KMT2A-rearranged lines in DepMap, lineage-split and reported as raw effect?
also called CLEC12A DepMap scan
Before doing any bench work, look in a public database that has already deleted every gene in hundreds of cancer cell lines, and ask whether deleting CLEC12A hurts KMT2A leukaemia cells in particular. It costs nothing and could settle the question.
The same thing, for a scientist
**Status: planned, not run.** Nothing in this graph currently reports a DepMap CLEC12A number and this entry must not be read as if it did. The analysis asks whether CLEC12A's CRISPR (Chronos) effect is more negative in KMT2A-rearranged lines than in comparator leukaemia lines. **Three corrections this project has already had to make to its own instrument apply in full and are written into the design**: the KMT2A-r arm must be **lineage-split** (it is myeloid-enriched, roughly 7 myeloid to 4 lymphoid, so an unsplit difference mixes fusion status with lineage); a **z is not an effect size** and the raw index-minus-comparator difference must be reported beside it; and **floor compression** makes a small difference uninterpretable for a gene already depleted in nearly all lines. For CLEC12A specifically there is a fourth, prior question: **is CLEC12A expressed at all in the lines being scored?** A CRISPR knockout of an unexpressed gene has no effect by construction, and a flat profile would then say nothing about dependency. The expression filter must be applied and reported before the dependency number is read.
Why a flat result would still be worth having
The falsifier in hyp-clec12a-is-a-dependency-not-only-a-marker-in-kmt2a-r is deliberately written so that a flat DepMap profile kills the cell-intrinsic version of the hypothesis before any bench money is spent. That is the cheapest falsification available on this beat, and it is the reason this analysis is written up rather than left implicit.
But it can only kill the hypothesis if the expression filter passes. A gene that is not expressed in the scored lines produces a flat profile whether or not it is a dependency in patient blasts - a clean, round, confident negative of exactly the shape this project's brief warns about. Report the expression distribution in the same table as the dependency number, or the negative is worthless.
| script | analyses/clec12a-dependency-kmt2a-r/run.py |
|---|---|
| status | planned |
| result summary | NOT RUN as of 2026-09-20. No CLEC12A DepMap number exists anywhere in this graph; do not quote one from this entry. |
inputs
[ "DepMap CRISPR (Chronos) gene effect", "DepMap expression (log2 TPM+1) for CLEC12A", "DepMap model annotations: lineage and KMT2A fusion status" ]
required outputs
[ "CLEC12A expression distribution across scored lines, reported FIRST", "raw index-minus-comparator Chronos difference, reported beside any z", "KMT2A-r arm split into myeloid and lymphoid halves, both reported", "fraction of all lines depleted, to check for floor compression" ]
- **The failure mode of this analysis is a confident flat negative.** An unexpressed gene and a non-dependency gene produce the same number. The expression filter is not optional.
- **A cell line has no marrow.** If CLEC12A's requirement is microenvironmental - NET DNA, myeloid neighbours - DepMap cannot see it at all, and a negative here would not licence dropping the hypothesis, only dropping the cell-intrinsic form of it.
- How many DepMap lines express CLEC12A above a defensible threshold at all? If the answer is very few, this analysis cannot be run and that itself should be recorded as a bounded negative rather than as a flat dependency profile.
- Is there a KMT2A-rearranged B-lineage line in DepMap that expresses CLEC12A? If not, the index disease is untestable by this route and only the bench experiment can answer it.
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
The result that motivates this analysis - that Clec12a knockout raises apoptosis, lowers colony number, reduces engraftment and prolongs survival in a NUP98::NSD1 model - was obtained in a murine genetic model rather than in a human cell-line dependency screen, so a public human screen is an independent test of it.
Clec12a is required for the pathogenesis of NUP98::NSD1 AML. · Mohanty S, Charles Cano F, Gabdoulline R, Lai CK, Othman B, Sudarsanam H, et al. · Blood Adv 2025What this evidence cannot show
Abstract read in full from raw NCBI efetch XML on 2026-09-20; full text not read. The murine model and a human cell-line panel differ in species, in genetic background and in the presence of a cooperating NRAS-G12D, so a disagreement between them would not straightforwardly refute either.
Evidence that argues against this (1)
- DepMap cell lines are long-established and heavily selected in vitro, and a receptor whose ligand is extracellular NET DNA has no obvious ligand in a culture flask. A flat DepMap profile is therefore a weaker refutation for THIS gene than for a typical cell-autonomous target, and the analysis should say so rather than over-reading its own negative. Malamud M et al. 2024
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset
A DepMap dependency number in this disease is not interpretable until the KMT2A-rearranged arm is split by lineage, because that arm is myeloid-enriched at roughly 7 myeloid to 4 lymphoid, and a z-scaled statistic in the depleted tail ranks narrowness rather than effect size.
Standing instrument corrections for DepMap analyses in this project (QUEEN_BRIEF.md, cycles 41 and 45) · Oncofusion knowledge graph coordinator · Oncofusion knowledge graph (docs/QUEEN_BRIEF.md) 2026What this evidence cannot show
This is a project-internal methodological record, not a peer-reviewed source. It is cited because the corrections are binding on this analysis's design; the underlying measurements (CEBPA and SPI1 as lineage artefacts, Spearman -0.92 between places gained and comparator SD, MEN1 2nd on z and 1st of 8,269 on raw effect) live in that document and were not re-derived by this beat.
Evidence that argues against this (1)
- A methodological rule recorded in a project's own brief can itself be wrong or out of date; the lineage composition of the DepMap KMT2A-r arm changes with each release, so the 7:4 split should be re-counted at run time rather than assumed. Oncofusion knowledge graph coordinator 2026
cheapest available falsification, planned not run
the gene whose dependency profile the planned scan would read; expression filter must be reported first
independent human cell-line test of the murine NUP98::NSD1 knockout result