CDC73 / parafibromin — the PAF1C subunit that was genetically tested, and a second tumour suppressor required by the fusion

This is a prediction, not a measurement. No one has seen this protein in this shape. The colours say how much the software trusts itself, and that is worth reading before anything else — the orange stretches are places the model is effectively saying this part has no fixed shape.

• Essential-function toxicity: CDC73 is required by 98.7% of cell lines. Degrading or inhibiting the protein is not a therapeutic option.
• Tumour-suppressor risk: CDC73 is a germline tumour suppressor. Chronic partial inhibition in a child raises the same unanswered second-cancer question this graph has already raised for menin, and with even less data.
• The validated reagent is a dominant-negative fragment, so the entire selectivity argument for MLL-PAFc rests on a tool that cannot be given to a patient.

1. Why are two parathyroid tumour suppressors, menin and parafibromin, both required by KMT2A-fusion leukemia? If this is not coincidence it implies a shared restraining function that the fusion subverts, and nobody appears to have asked the question.
2. Which PAF1C subunit makes the direct contact with the MLL CxxC-flanking region? The interface is genetically validated as a selective target but its subunit-level geometry is undefined, which blocks any structure-based campaign.
3. Does menin inhibition alter PAFc occupancy at fusion target loci, or vice versa? Two selective interfaces on the same fusion protein are the natural orthogonal combination this project has been looking for, and their interaction is untested.

1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   Conditional Cdc73 deletion blocks AML cell proliferation and establishes the PAF1 complex as required in MLL-fusion AML, while targeted disruption of the MLL-PAFc interaction selectively inhibits MLL-fusion-driven AML without impairing normal haematopoietic stem cell function.

   MLL fusion protein-driven AML is selectively inhibited by targeted disruption of the MLL-PAFc interaction. · Muntean AG, Chen W, Jones M, Granowicz EM, Maillard I, Hess JL · Blood 2013
   PMID 23900238DOI
   ▸What this evidence cannot show

   Mouse AML models and a dominant-negative protein fragment, not infant KMT2A::AFF1 B-ALL and not a drug. The paper states PAFc is required indiscriminately for AML proliferation, so CDC73 loss is not a selective intervention; only the interface is.

   ▸Evidence that argues against this (1)

   • CDC73/parafibromin is a bona fide germline tumour suppressor, which is the reason a CDC73-directed intervention cannot be pursued as written. Thirteen distinct heterozygous germline inactivating mutations were identified in fourteen families with hyperparathyroidism-jaw tumour syndrome, and three somatic inactivating mutations were found among 48 parathyroid adenomas, none in normal controls. The entry acknowledges this framing, but the constraint deserves a resolvable citation attached to the evidence item that asserts the dependency: a compound that reduces CDC73 function is a compound that phenocopies a cancer predisposition allele. Carpten JD et al. 2002
2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   The PAF complex regulates Prmt5 to facilitate the progression and maintenance of MLL-fusion leukemia.

   The PAF complex regulation of Prmt5 facilitates the progression and maintenance of MLL fusion leukemia. · Serio J, Ropa J, Chen W, Mysliwski M, Saha N, Chen L, Wang J, Miao H, Cierpicki T, Grembecka J, Muntean AG · Oncogene 2018
   PMID 28945229DOI
   ▸What this evidence cannot show

   This paper concerns PRMT5, which is a different enzyme from PRMT1; the two are easily conflated in this literature and should not be. Mouse and cell-line models.

   ▸Evidence that argues against this (1)

   • Read this evidence item with the disambiguation that PRMT5 is not PRMT1. The PAF-complex regulation described here is of PRMT5, a type II arginine methyltransferase with a different substrate specificity and a separate inhibitor pipeline, and the same paper is carried in gene-prmt1 E3 explicitly as a disambiguation warning because the two enzymes are routinely conflated in the MLL literature. Anyone building a CDC73-to-arginine-methylation argument from this item should not carry it across to PRMT1, whose case in MLL-fusion leukemia rests on MLL-EEN and on FLT3 methylation, both different mechanisms. Zhu Y et al. 2019
3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   PAF1 complex interactions with SETDB1 mediate promoter H3K9 methylation and transcriptional repression of HOXA9 and MEIS1.

   PAF1 complex interactions with SETDB1 mediate promoter H3K9 methylation and transcriptional repression of Hoxa9 and Meis1 in acute myeloid leukemia. · Ropa J, Saha N, Chen Z, Serio J, Chen W, Mellacheruvu D, Zhao L, Basrur V, Nesvizhskii AI, Muntean AG · Oncotarget 2018
   PMID 29774127DOI
   ▸What this evidence cannot show

   Oncotarget, weaker editorial reputation. Demonstrated in AML models; the relative weight of the repressive arm in B-ALL is unknown.

   ▸Evidence that argues against this (1)

   • This is PAF1-complex biology acting to REPRESS HOXA9 and MEIS1 through SETDB1, which is the opposite direction from the fusion's programme. An entry placing CDC73 in a fusion interactome has to say which role it is claiming. Cell Reports 2014
4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

   CDC73 is DepMap common-essential with mean Chronos -1.122 across 1,258 lines and depletion in 98.7% of lines; gnomAD reports pLI ~1.0 and LOEUF 0.361 with 21 observed against 83.7 expected loss-of-function variants; HPA reports low tissue specificity with bone marrow 8.4 nTPM and 0.5 nTPM in classical monocytes and naive B cells.

   DepMap CRISPR (Chronos) gene effect via the Open Targets Platform GraphQL API; gnomAD v4 constraint via the gnomAD GraphQL API; Human Protein Atlas consensus tissue nTPM via the HPA API · Broad Institute DepMap; gnomAD; Human Protein Atlas; Open Targets Platform · Databases 2026
   ▸What this evidence cannot show

   DepMap release version not stated by the API. Strong LoF constraint in a gene that is also a classical tumour suppressor is expected and does not by itself indicate intolerance to partial pharmacological inhibition.

   ▸Evidence that argues against this (1)

   • CTR9 is a germline tumour suppressor in a childhood cancer, which is a hard constraint on treating it as a target in paediatric leukemia. Inactivating germline CTR9 mutations were found in 3 of 35 Wilms tumour families by exome and Sanger sequencing and in none of 1,000 population controls (P < 0.0001); each mutation segregated with disease and a second mutational event was present in the available tumours, establishing CTR9 as a Wilms tumour predisposition gene acting as a tumour suppressor. A second family with a novel splice-acceptor CTR9 germline variant and tumour loss of heterozygosity was reported subsequently (PMID 29292210). Whatever CTR9 contributes to the MLL-PAFc interaction, pharmacologically reducing CTR9 function in a child is reducing a tumour suppressor. Hanks S et al. 2014
5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

   Across 54 DepMap leukemia lines CDC73 correlates with MEN1 at r = -0.076, confidence interval -0.306 to 0.150, crossing zero.

   Leukemia-restricted co-dependency with MEN1, bootstrap and permutation testing; analyses/wrad-discrimination/ in this project · Broad Institute DepMap; Open Targets Platform; analysis run by this project · Database and derived analysis 2026
   ▸What this evidence cannot show

   Our own computation in 54 lines. A pan-essential gene has no variance to correlate, so this null is uninformative about biology.

   ▸Evidence that argues against this (1)

   • A correlation of −0.076 with an interval from −0.306 to 0.150 is a null result. The entry records it correctly and the surrounding dossier should not be read as if a co-dependency had been found. Oncofusion knowledge graph (internal analysis) 2026
6. E6established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   Two high-resolution crystal structures of the CDC73 N-terminal region exist, 5YDE (residues 1-111, 1.02 A) and 5YDF (residues 1-100, 1.40 A), and CDC73 is additionally resolved within Pol II elongation-complex cryo-EM structures.

   RCSB PDB entries 5YDE and 5YDF, verified against the RCSB data API for title, experimental method and resolution · RCSB Protein Data Bank · RCSB PDB 2026
   ▸What this evidence cannot show

   These entries cover only the first ~110 residues of a 531-residue protein; the bulk of CDC73 including any MLL-proximal surface is not covered by an isolated high-resolution structure.

   ▸Evidence that argues against this (1)

   • Counting PDB entries measures how tractable a protein was to CRYSTALLISE, which correlates with being small, ordered and well-behaved rather than with being druggable. What decides design value is whether the deposited density covers the site that matters, at a resolution that supports modelling, in the relevant state — and this graph has repeatedly found structures described in the direction that overstates what is resolved. 1.02 A and 1.40 A on the N-terminal 111 residues is superb data on a small fragment; CDC73 being otherwise resolved only inside cryo-EM elongation complexes means the rest of it has no independent structure. RCSB PDB 2026
7. E_q95_cdc73_loc1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   UniProt Q6P1J9 carries exactly ONE subcellular location value for parafibromin - 'Nucleus', on three ECO:0000269 citations, all of them 2005 papers about PAF1-complex membership and cyclin D1 regulation - together with a mapped bipartite nuclear localization signal at residues 125-139, and it carries NO nucleolar term at all, although three nucleolar localization signals have been mapped in the protein and the nucleolar pool is separately measurable in human tumours.

   UniProtKB Q6P1J9 (CDC73_HUMAN), SUBCELLULAR LOCATION comment, evidence codes and feature table, fetched live 2026-09-22 · UniProt Consortium · Databases 2026
   ▸What this evidence cannot show

   This is a statement about the database record, not about the biology, and the direction of the defect is the OPPOSITE of the one this project usually finds. The three cited papers are real parafibromin papers and none of them is off-topic in the way UniProt's KMT2A or AFF4 citations are; what is missing is precision, not honesty. The feature table does carry the NLS, which is more than MEN1's record does.

   ▸Evidence that argues against this (1)

   • ARGUING AGAINST TREATING THIS AS A DATABASE ERROR: one of the three cited papers is the paper that placed parafibromin in the human PAF1 complex, so 'Nucleus' is exactly the claim those papers support and a curator who wrote only what the cited experiments showed would write only 'Nucleus'. The sharper point is comparative. On CTR9, one subunit away in the same complex, UniProt's single value is the MORE precise 'Nucleus speckle' resting on a source whose own words are 'predominantly in the nucleus' - precision the record gained in transit. Here the record LOST precision that the literature has. Two subunits of one complex, two opposite failures, and only reading the cited papers separates them; no evidence-code filter distinguishes the two cases, because both are ECO:0000269. Rozenblatt-Rosen O et al. 2005
8. E_q95_cdc73_loc2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   Parafibromin's subnuclear address is the NUCLEOLUS, and it is determined by three nucleolar localization signals at residues 76-92, 192-194 and 393-409 that are clusters of arginine and lysine; the bipartite NLS at 125-139 is the primary determinant of NUCLEAR localisation and does not mediate nucleolar localisation, while the three NoLSs play only a minor role in nuclear localisation and are critical for the nucleolar pool - so the two compartments are separately addressed in one protein.

   Nucleolar localization of parafibromin is mediated by three nucleolar localization signals. · Hahn MA, Marsh DJ · FEBS Lett 2007
   PMID 17923126DOI
   ▸What this evidence cannot show

   Signal mapping by mutagenesis of transfected constructs is a statement about DETERMINANTS rather than about where the endogenous protein sits, and the abstract does not name the cell line. Independent mapping of three NoLS mutations of CDC73 reaches the same three-signal architecture (PMID 24340015), and a separate missense mutant characterisation adds a fourth (PMID 24842573), so the determinant map is replicated; the endogenous confirmation comes from human tumour tissue and is recorded separately as this entry's nucleolar-loss evidence.

   ▸Evidence that argues against this (1)

   • THIS CONTRADICTS THE ONLY OTHER SUBNUCLEAR MEASUREMENT OF THIS COMPLEX IN THIS GRAPH, AND THE CONTRADICTION IS THE INTERESTING PART. Tranche 5 recorded, on gene-ctr9, that in budding yeast all five wild-type Paf1C components - Paf1, Ctr9, Cdc73, Rtf1 and Leo1 - are present in the NUCLEOPLASM BUT NOT THE NUCLEOLUS, and that pulling the complex off Pol II by paf1 or rtf1 loss is what ADMITS it to the nucleolus. Human parafibromin instead carries three constitutive nucleolar addressing signals. The two results are reconcilable in exactly one way that is testable: the nucleolar parafibromin pool is the pool NOT in Paf1C. The yeast paper measured the complex, the NoLS papers measured the subunit, and the experiment nobody has done is to ask whether nucleolar parafibromin co-purifies with PAF1, CTR9 and LEO1 or with none of them. Until that is done, this graph should not write 'PAF1C is nucleolar'. Porter SE et al. 2005
9. E_q95_cdc73_loc3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   The nucleolar pool of parafibromin is separately lost in human disease tissue: nucleolar parafibromin immunoreactivity was ABSENT in three of 23 parathyroid carcinomas and one of 16 atypical adenomas scored by high-power microscopy in 82 tumours, all three carcinomas carried HRPT2/CDC73-inactivating mutations predicted to abolish the three nucleolar localization signals, and in those tumours NUCLEAR parafibromin was still expressed - so the nucleolar compartment can be lost while the nuclear compartment is retained.

   Absence of nucleolar parafibromin immunoreactivity in subsets of parathyroid malignant tumours. · Juhlin CC, Haglund F, Obara T, Arnold A, Larsson C, Höög A · Virchows Arch 2011
   PMID 21221636DOI
   ▸What this evidence cannot show

   Immunohistochemistry on parathyroid tumours, n = 4 positive events out of 82 tumours, so the event is rare and the study is an antibody-scoring study rather than a mechanistic one; brief rule 74 applies in that the clone determines what could be seen, and the abstract does not name it. The germline-mutant arm supplies the mechanism independently (PMID 20304979): a HPT-JT L95P missense mutant is markedly deficient in nucleolar localisation, still assembles with endogenous PAF1 complex components by co-immunoprecipitation, and acquires a growth-stimulatory dominant-interfering phenotype in NIH-3T3 and HEK293.

   ▸Evidence that argues against this (1)

   • THE SYSTEM IS WRONG FOR THIS GRAPH AND THAT HAS TO BE SAID IN THE CLAIM, NOT ONLY HERE (brief rule 73): every endogenous measurement of nucleolar parafibromin is in PARATHYROID tissue, in an endocrine tumour syndrome, and the mechanistic arm is in NIH-3T3 fibroblasts and HEK293. There is no haematopoietic cell of any kind in the nucleolar literature for this protein. The countervailing consideration is that it is the same allele series in both arms - mutations abolishing the same three NoLS - which is a stronger form of agreement than two independent systems reporting the same compartment, because the genotype is doing the work. But a nucleolar tumour-suppressor function established in parathyroid cells cannot be assumed in a KMT2A-rearranged blast, where the nucleolar/ribosome-biogenesis module is the single most enriched complex in the fusion interactome and the nucleolar biology may therefore be under completely different control. Panicker LM et al. 2010
10. E_q95_cdc73_loc4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    CDC73's cell type and cell state are measured, and in mouse rather than human: CDC73 is required for both fetal and adult haematopoiesis, loss of Cdc73 in haematopoietic cells is lethal through extensive bone marrow failure, it has an essential cell-autonomous role for adult haematopoietic stem cell function in vivo, its deletion causes cell-cycle defects in haematopoietic progenitors, and expression profiling shows DIFFERENTIAL regulation of the Hoxa9/Meis1 programme in progenitors compared with AML cells - a leukaemia-specific regulation of self-renewal genes inside a gene the normal stem cell also needs.

    The PAF1c Subunit CDC73 Is Required for Mouse Hematopoietic Stem Cell Maintenance but Displays Leukemia-Specific Gene Regulation. · Saha N, Ropa J, Chen L, Hu H, Mysliwski M, Friedman A, Maillard I, Muntean AG · Stem Cell Reports 2019
    PMID 31031188DOI
    ▸What this evidence cannot show

    Mouse conditional knockout, and the leukaemia arm is MLL-fusion AML rather than infant B-lineage KMT2A::AFF1, so both halves of the result are imported. No compartment was measured in either arm - this is a requirement and gene-expression result, and it is recorded under cell type and cell state rather than localisation for that reason.

    ▸Evidence that argues against this (1)

    • THIS RESULT ARGUES AGAINST TARGETING CDC73 AND THE ENTRY SHOULD NOT READ AS THOUGH A LEUKAEMIA-SPECIFIC GENE PROGRAMME IMPLIED A THERAPEUTIC WINDOW. The same paper that reports leukaemia-specific regulation reports that Cdc73 loss is LETHAL through bone marrow failure and is cell-autonomously required by the adult HSC, and DepMap has CDC73 depleted in 98.7% of 1,258 lines at a mean Chronos of -1.12, which is among the most floor-compressed values in this graph. So the 'window' is a difference in which genes move, in a cell that dies either way. The germline data point the same direction: CDC73 is a tumour-suppressor gene in humans, so reducing its function pharmacologically in a child is reducing a tumour suppressor - the identical objection tranche 5 recorded for CTR9 and Wilms tumour, now the second PAF1C subunit in this graph to carry it. Juhlin CC et al. 2011
11. E_q95_cdc73_loc5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    Bounded 2026-09-22: no localisation measurement of CDC73 exists in any KMT2A-fusion context. The gene clause over CDC73, HRPT2 and parafibromin returns 656 records and the nine-term KMT2A/MLL fusion clause intersects it at 4; the gene clause with a localisation clause returns 100 and with the fusion clause added returns 0; the gene clause with a subnuclear clause returns 10 and with the fusion clause added returns 0. The nucleolar sub-query returns 8 records of which 6 are the parafibromin nucleolar literature itself and none is in a leukaemic cell, and the speckle sub-query returns 0.

    NCBI E-utilities esearch db=pubmed; every count stored with the querytranslation read back, in data.localization_bounded_absence · NCBI E-utilities · Databases 2026
    ▸What this evidence cannot show

    Titles and abstracts only; supporting figures were not searched, which matters here because brief rule 80's worked example is exactly a compartment measurement living in a figure panel. One tag per term and every count stored with its querytranslation. Rule 64 was applied and changes the reading: the complex form is what reaches this protein's real subnuclear literature at all - the yeast Paf1C paper that measures Cdc73 has no CDC73 symbol in its title - so the zero bounds the SUBUNIT-plus-fusion literature and not the question.

    ▸Evidence that argues against this (1)

    • THE ZERO IS WEAKER THAN IT LOOKS IN ONE DIRECTION AND STRONGER IN ANOTHER, AND BOTH SHOULD BE STATED. Weaker: CDC73 is the PAF1C subunit that HAS been genetically tested in MLL-fusion AML, so a fusion-context zero on localisation sits inside a literature that is otherwise active on this exact gene and disease - the absence is of a compartment measurement, not of interest. Stronger: the nucleolar sub-query is the one query in this beat where the contaminated-token hazard the registry records for 'nucleolar' did NOT fire - 6 of 8 hits are genuinely about parafibromin nucleolar localisation, against 2 of 2 off-topic for USP8 and roughly 19 of 20 off-topic for CDK4 in this same tranche. A gene with its own real nucleolar literature is protected from that contamination by having enough true positives to see. Pazienza V et al. 2013

• part of →PAF1 — pan-essential subunit of a complex the fusion needs, where the window is in the interface and not in the protein

  Counter-evidence: CDC73/parafibromin is a germline tumour suppressor in HPT-JT syndrome

• part of →PAF1 — pan-essential subunit of a complex the fusion needs, where the window is in the interface and not in the protein

  PAF1C subunit; the one genetically deleted to test the complex in MLL-fusion AML

• related to →MEN1 / menin — the adaptor the fusion cannot work without

  both are parathyroid tumour suppressors that KMT2A-fusion leukemia nonetheless requires

• regulates →H3K9me3 - measured in KMT2A::AFF1 cells, in a supplementary figure, and it rises GENERALLY where H3K27ac falls SPECIFICALLY in the same experiment

  the CDC73_3YF mutant stabilises the PAF1c-SETDB1 interaction and raises the mark

• ← depends onPRMT5 - the type II arginine methyltransferase this graph has reached for sixteen times, whose mark on H3R2 RECRUITS WDR5 to chromatin

  PRMT5 transcription is directly activated by the PAF complex in MLL-AF9 AML

• ← related toPRMT3 - the mostly-cytoplasmic ribosome methyltransferase that also writes the ACTIVATING mark on the residue PRMT5 represses, has an acute-leukaemia degrader, and is invisible to this project's own mark-query rule

  PRMT3 transcript falls on CDC73 excision in MLL-AF9 cells - a full-text-only datum

• regulates →H4R3me2s - the repressive symmetric arginine mark that recruits DNMT3A, blocks a KMT2 enzyme's own PHD module, and is the ONLY arginine mark ever measured in a KMT2A-fusion system

  conditional Cdc73 excision in murine MLL-AF9 REDUCES PRMT5-mediated H4R3me2s - the only arginine-mark measurement in any KMT2A-fusion system, and it is a writer-abundance result

• recruits →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

  parafibromin aa 128-227 recruits SUV39H1 to the cyclin D1 promoter for H3K9 methylation; SHP2 dephosphorylation flips it to beta-catenin

• recruits →SETDB1 (Q15047, KMT1E/ESET) - the second H3K9me3 writer: it represses the fusion's own targets via the PAF1 complex, it is required in fetal liver HSPCs and in leukaemic stem cells, and it is a broad dependency in half the CRISPR panel - three published directions that do not agree

  PAF1c interacts with SETDB1 in MLL-AF9 cells; the CDC73_3YF mutant stabilises the interaction and cannot support AML growth

• related to →The nucleolus — the one compartment with direct published images of a KMT2A fusion inside it

  CDC73/parafibromin is the best-evidenced NUCLEOLAR protein in this graph's PAF1C set: three mapped nucleolar localization signals at 76-92, 192-194 and 393-409, separate from the bipartite NLS at 125-139 (PMID 17923126), and the nucleolar pool is lost in human parathyroid carcinomas whose HRPT2 mutations abolish those signals while nuclear staining is retained (PMID 21221636). No leukaemic cell has been imaged (Q95 2026-09-22)

• related to →PAF1 is annotated as punctate everywhere in the nucleus EXCEPT the nucleolus - and a 2024 proximity map puts it in the nucleolus

  CDC73 SHARPENS this node's contested status and does not resolve it. Human parafibromin carries three constitutive NoLS; yeast wild-type Paf1C including Cdc73 is nucleolus-EXCLUDED and enters the nucleolus only when pulled off Pol II (PMID 15643076). The testable reconciliation is that the nucleolar parafibromin pool is the pool NOT in PAF1C - nobody has asked whether nucleolar parafibromin co-purifies with PAF1, CTR9 and LEO1 (Q95 2026-09-22)

• related to →CTR9 — a PAF1C subunit the field assumes is a fusion partner, with no paper that tested it

  Two subunits of one complex with OPPOSITE database failures, and only opening the cited papers separates them: CTR9's single UniProt value is 'Nucleus speckle' resting on a source that says only 'predominantly in the nucleus' - precision gained in transit - while CDC73's single value is a bare 'Nucleus' although three nucleolar signals are mapped in the protein. Both are ECO:0000269; no evidence-code filter tells them apart (Q95 2026-09-22)

• expressed in →Bone marrow haematopoietic stem cell — the postnatal comparator, and why the same fusion behaves differently here

  CDC73 is cell-autonomously required for adult haematopoietic stem cell function in vivo and for fetal haematopoiesis; loss in haematopoietic cells is lethal through bone marrow failure, with cell-cycle defects in progenitors (PMID 31031188, mouse conditional knockout). A requirement, not an image (Q95 2026-09-22)

• regulates →The stem-like self-renewal state — a limited stem-cell module grafted onto a committed cell

  CDC73 regulates the Hoxa9/Meis1 self-renewal programme DIFFERENTLY in normal haematopoietic progenitors and in MLL-fusion AML cells - the clearest normal-versus-leukaemic separation available for any PAF1C subunit (PMID 31031188, mouse). The window is in which genes move, not in whether the cell survives (Q95 2026-09-22)

• ← testsAuditing the seven strikes before applying them: nine of ten decisions stand, four for different reasons than were given, and one clinical argument fails

  strike confirmed on pan-essentiality; HPT-JT reasoning demoted for the same reason