Oncofusion

start / graph / gene-psat1

Genemoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q109-serine-one-carbon

PSAT1 (Q9Y617) - a DIRECT ATF4 target, and the gene that shows the brief's menin-selectivity claim was PHGDH's and not the pathway's

also called Phosphoserine aminotransferase · SERC_HUMAN · PSAT · EPIP

PSAT1 does the second step of serine manufacture, and a menin inhibitor drops it about five-fold in KMT2A-rearranged leukaemia cells. Unlike PHGDH, a DOT1L inhibitor reproduces most of that drop - which fits PSAT1 being a direct target of the stress factor ATF4 rather than of menin itself.

The same thing, for a scientist

PSAT1 (Q9Y617, 370 aa, exactly one reviewed human record) catalyses the second, PLP-dependent step of phosphorylated serine biosynthesis and additionally acts as an inhibitor of ferroptosis [E5]. **Measured here:** RS4;11 day 7, VTP-50469 330 nM, **PSAT1 -2.40 log2** from a baseline of 8.35 log2 CPM - and **EPZ5676 1 uM gives -1.85, reproducing 77% of the effect, for a VTP-minus-EPZ margin of only -0.55** [E1]. **That number corrects a premise this beat was dispatched on.** The dispatch brief paired "PHGDH is -3.24 under VTP-50469 and only -1.52 under EPZ5676" with "and PSAT1 is -2.40", which reads as though PSAT1 were part of the same menin-selective evidence. It is not: -2.40 is PSAT1's VTP value, its margin is a third of PHGDH's, and on compound-selectivity PSAT1 sits between the ISR-output arm (-0.02) and PHGDH (-1.72). **The mechanism fits.** ATF4 directly binds the PSAT1 promoter in AML, driving glutathione synthesis, suppressing ferroptosis and causing venetoclax resistance [E3]; and PSAT1 is the gene through which lycorine kills **B-cell** ALL lines, including REH [E4]. So PSAT1 looks ATF4-proximal where PHGDH looks chromatin-proximal, and the two enzymes of one pathway are reached by different routes. DepMap: mean **-0.2531 across 1,258 screens, 8.2% below -0.5, leukaemia -0.2669** - eight times PHGDH's fraction and still not a dependency [E2]. Germline biallelic loss causes phosphoserine aminotransferase deficiency (intractable seizures, acquired microcephaly, hypertonia) and, at the severe end, **Neu-Laxova syndrome 2** [E5][E6].

What it looks like
loading 3E77…
PDB 3E77 ↗
shape

This is real experimental data — atoms placed by crystallography or cryo-EM, deposited by the people who did the experiment.

I have never read one of these pictures before

A protein is one long chain of amino acids that folds up. The ribbon traces that chain from one end to the other; it is not a wire, it is a simplification of where the backbone runs. Drag to rotate, scroll to zoom.

Domains are the stretches of chain that do a particular job, and they are what the coloured blocks above the picture mark out. Click one and the view flies to it. In this project the important question about any domain is usually whether the fusion still has it — the labels say lost where it does not.

Anything in yellow sticks is a bound small molecule, which is usually the drug.

Go deeper
The premise correction, stated plainly

The dispatch brief's third premise was: "PHGDH is -3.24 under VTP-50469 and only -1.52 under EPZ5676, and PSAT1 is -2.40. So the ISR collapse is compound-independent while the serine collapse is partly menin-selective."

The conclusion survives. The supporting pair does not, because the two numbers are not the same kind of number 1:

geneVTP-50469EPZ5676marginEPZ reproduces
PHGDH-3.24-1.52-1.7247%
PSAT1-2.40-1.85-0.5577%
CBS-4.07-2.60-1.4764%
GPT2-1.54-0.51-1.0333%
ALDH1L2-3.40-2.91-0.4985%
MTHFD2-1.16-1.13-0.0397%
ASNS-1.72-1.47-0.2586%

The menin-selective signal is concentrated in PHGDH, CBS and GPT2, not spread across the pathway. MTHFD2 and ASNS are as compound-independent as the ISR genes. A reader who takes "the serine arm is menin-selective" as a uniform property of the pathway will be wrong about more than half of its members, and the entry for each one now carries its own margin so the claim cannot be made at pathway level again.

REH is the wild-type discriminator and it is not a clean one for PSAT1

On the array, both PSAT1 probes are expressed in REH (8.59 and 10.65 log2 RMA, floor 6.0) and move positive under MI-2-2, +1.18 and +0.49, where RS4;11 gives -0.77 and -0.49 and SEM -0.96 and -0.34 1. Direction is right for a fusion-dependent programme.

But PSAT1 matters in REH. Lycorine inhibits growth and induces apoptosis in REH and NALM-6, and RNA-seq, ectopic expression, shRNA knockdown and LC-MS all converge on PSAT1 as the mediator, with synergy against cytarabine and activity in a cell-line-derived xenograft 4. So "flat in the KMT2A-wild-type line under a menin inhibitor" means the fusion programme does not regulate PSAT1 there; it does not mean the wild-type line does not need PSAT1. Those are different claims and this graph has conflated them before.

Identifier traps measured on this symbol

gene_exact:PSAT1 AND organism_id:9606 AND reviewed:true returns exactly one reviewed record, Q9Y617 / SERC_HUMAN, 370 aa, gene_primary read back as PSAT1, with the single UniProt synonym PSA 5.

That synonym is the trap, and it is a UniProt-listed one - the same shape as NMP22 for NUMA1. Measured 2026-09-22 with one tag per term 7:

operandrecordsfusion-scoped
PSAT1[tiab]3521 (PMID 29672864)
PSAT[tiab]3270
EPIP[tiab]620
PSA[tiab]51,9945

PSA is prostate-specific antigen (KLK3) in essentially all of that literature. Excluded, not unioned, and the exclusion is recorded here so a later beat does not rediscover it. Note that the exclusion is a judgement about the token's referent, not about its size: PSAT[tiab] is nearly as large as PSAT1[tiab] and is the same object.

Facts and figures
gene symbolPSAT1
uniprotQ9Y617
length AA370
entry nameSERC_HUMAN
protein namePhosphoserine aminotransferase
ensembl gene IDENSG00000135069
PDB noteall three PDB cross-references of Q9Y617 are listed; NONE was fetched, and no chain, resolution or residue range is claimed from them
accession resolutiongene_exact:PSAT1 AND organism_id:9606 AND reviewed:true returns EXACTLY ONE reviewed row (rule 62 asserted): Q9Y617, SERC_HUMAN, gene_primary PSAT1, synonym PSA, 370 aa.
localizationNOT ANNOTATED IN UNIPROT - a bounded absence. Q9Y617 carries no SUBCELLULAR LOCATION comment; the comment types present are ACTIVITY REGULATION, ALTERNATIVE PRODUCTS, BIOPHYSICOCHEMICAL PROPERTIES, CATALYTIC ACTIVITY, COFACTOR, DISEASE, FUNCTION, INTERACTION, PATHWAY, PTM, SIMILARITY, SUBUNIT and TISSUE SPECIFICITY. PHGDH (O43175) and GPT2 (Q8TD30) are unannotated the same way. No localisation experiment on PSAT1 has been performed in any KMT2A-rearranged cell.
subcellular localizationUNKNOWN from UniProt. Cytosolic by pathway inference only, recorded as an inference and not as a fact.
subnuclear localizationNONE, and answered negative rather than left blank: PSAT1 has no reported nuclear pool. SHMT1 and SHMT2 in the same panel DO carry ECO:0000269 Nucleus statements, so a nuclear pool in one-carbon metabolism is not implausible and PSAT1's absence from that list is informative rather than merely unrecorded.
cell typeRS4;11 and SEM (KMT2A::AFF1 B-ALL), REH (KMT2A wild type, ETV6::RUNX1 B-ALL), MOLM13 (KMT2A::MLLT3 AML) and MV4;11 (KMT2A::AFF1 myeloid) measured here. REH and NALM-6, both KMT2A wild type, in the lycorine B-ALL study [E4]. AML lines and relapsed patient samples in the venetoclax-resistance study [E3]. Ewing sarcoma in PMID 29672864. NO primary infant KMT2A-rearranged sample and NO normal human pro-B cell.
cell stateMeasured in the drug-treated, arresting state at day 7 of continuous menin-inhibitor exposure, under uncontrolled medium serine supply - no menin-inhibitor experiment in any of the three series reports the amino-acid composition of the medium, which is a live alternative explanation for an amino-acid-response readout. At day 2 PSAT1 carries 44% of its day-7 fall (-1.05 of -2.40), an early-and-progressive profile closer to the canonical menin targets (MEF2C 76%, PBX3 64%) than to the late ISR-output genes (TRIB3 13%, ATF5 22%). The other state in which PSAT1 is characterised is VENETOCLAX RESISTANCE: PSAT1 is up in resistant lines and relapsed samples and its knockdown restores sensitivity through ferroptosis [E3].
epigenetic readoutINDIRECT, and weaker than PHGDH's. PMID 29672864 reports that menin inhibition diminished expression of SSP genes including PSAT1 and reduced H3K4me3 enrichment at the PHGDH promoter - the H3K4me3 measurement is reported at PHGDH, not at PSAT1, so no chromatin mark has been measured at the PSAT1 locus under menin inhibition in any system. The transcriptional input demonstrated for PSAT1 is ATF4 binding at its promoter in AML [E3], which is transcription-factor occupancy rather than an epigenetic mark. NOT measured in any KMT2A-rearranged cell: no H3K4me3, H3K79me2, KMT2A or menin ChIP at the PSAT1 locus.
shown for fusionPARTLY. The transcript collapse is shown in KMT2A::AFF1 (RS4;11, SEM, MV4;11) and KMT2A::MLLT3 (MOLM13). The ATF4-at-the-PSAT1-promoter mechanism is shown in AML lines of unstated fusion status. The killing-through-PSAT1 result is shown in KMT2A WILD-TYPE B-ALL (REH, NALM-6), i.e. as a lineage result and not a fusion result. NOTHING places PSAT1 in a dependency relationship with any KMT2A fusion.
absence classmeasured-not-analysed
absence class noteThe compound-selectivity comparison existed in deposited data and had not been run. The never-measured items: a chromatin mark at the PSAT1 locus in any KMT2A-rearranged cell, and a PSAT1 dependency measurement in a KMT2A::AFF1 model.
PDB IDS
[
  "3E77",
  "8A5V",
  "8A5W"
]
the measurement rs411 d7 log2fc
{
  "vtp50469_330nM": -2.4,
  "epz5676_1uM": -1.85,
  "vtp_minus_epz": -0.55,
  "epz_fraction_of_vtp": 0.77,
  "dmso_baseline_log2cpm": 8.35,
  "detected": true,
  "floor_collision": false,
  "d2_vtp": -1.05,
  "d2_share_of_d7": 0.44,
  "molm13_d7_vtp": -1.61,
  "molm13_d7_epz": -1.19
}
the measurement array gse63664 every probe
{
  "220892_s_at": {
    "rs411_dmso_log2_rma": 10.35,
    "rs411_mi22": -0.77,
    "sem_dmso": 11.61,
    "sem_mi22": -0.96,
    "reh_dmso": 8.59,
    "reh_mi22": 1.18
  },
  "223062_s_at": {
    "rs411_dmso_log2_rma": 11.67,
    "rs411_mi22": -0.49,
    "sem_dmso": 12.07,
    "sem_mi22": -0.34,
    "reh_dmso": 10.65,
    "reh_mi22": 0.49
  },
  "detection_floor_used": "6.0 log2 RMA; both probes above it in all three lines, so every value here is a measurement"
}
mv411 independent lab
{
  "menin_vtp50469_500nM": -2.17,
  "dot1l_sgc0946_5uM": -1.19,
  "margin": -0.98,
  "dmso_baseline_log2cpm": 4.43,
  "caveat": "housekeeping is not flat in this matrix (ACTB -0.73), so magnitudes are inflated; the comparator is a different DOT1L compound at a different dose"
}
essentiality
{
  "depmap_release": null,
  "depmap_release_note": "Rule 83: the Open Targets payload carries no release identifier. Fetched 2026-09-22.",
  "depmap_screens": 1258,
  "depmap_tissue_blocks_flattened": 29,
  "chronos_mean": -0.2531,
  "chronos_median": -0.2431,
  "pct_lines_below_minus_0_5": 8.2,
  "leukaemia_screens": 57,
  "leukaemia_mean_gene_effect": -0.2669,
  "depmap_common_essential": "not asserted",
  "within_panel_comparison": "8.2% against PHGDH's 1.1% and PSPH's 0.6% - PSAT1 is the most required of the three canonical SSP enzymes, which is the opposite of the order their fold changes would suggest, and is a reason not to read a fold change as a dependency ranking",
  "assay_limitation": "serine-replete medium understates a serine-synthesis dependency"
}
How this could bite us back
  • **Germline biallelic PSAT1 loss causes Neu-Laxova syndrome 2 at the severe end and intractable seizures with acquired microcephaly and hypertonia at the milder end.** Like PHGDH, the phenotype is neurodevelopmental and intrauterine, and the patients here are infants. No PSAT1 inhibitor exists; the only reported pharmacological route is a plant alkaloid with many targets.
  • **PSAT1's fall under a menin inhibitor may be protective rather than exploitable.** PSAT1 suppresses ferroptosis and restrains JNK signalling; lowering it in AML restores venetoclax sensitivity. That is a reason to pair a menin inhibitor with venetoclax, and simultaneously a warning that lowering PSAT1 sensitises cells to lipid peroxidation, which is not a selective pressure - normal proliferating marrow has the same chemistry.
  • **The direction of the vulnerability is contested in the literature this entry cites.** One report makes high PSAT1 a resistance factor to be removed; a preprint makes PSAT1 SUPPRESSION the thing that creates a targetable auxotrophy. Both cannot be used to argue for the same intervention, and this entry does not resolve which applies to KMT2A-rearranged ALL.
  • **Do not read the REH null as evidence that wild-type B-ALL is spared.** REH is killed through PSAT1 by lycorine with cytarabine synergy and xenograft activity. A serine-pathway intervention in B-ALL has no demonstrated fusion selectivity at all.
Still open
  1. Does ATF4 occupy the PSAT1 promoter in a KMT2A-rearranged cell, and does that occupancy fall under a menin inhibitor? PMID 42044713 reports the binding in AML. One ChIP-qPCR in RS4;11 would convert PSAT1 from 'behaves like an ATF4 target' to 'is one' in this disease, and it is the same experiment as the PHGDH H3K4me3 ChIP on a different locus.
  2. Does menin inhibition sensitise KMT2A-rearranged cells to VENETOCLAX through PSAT1? PSAT1 drives venetoclax resistance by suppressing ferroptosis, and a menin inhibitor lowers PSAT1 2.40 log2. This graph holds drug-venetoclax and revumenib-venetoclax combinations are already in trials, so the prediction is testable in an existing clinical context - and it is a different and better-founded combination prediction than the asparaginase one.
  3. If PSAT1 suppression CREATES serine auxotrophy (PMID 40462988, preprint), does a menin inhibitor create a serine dependency rather than exploit one? That inverts the therapeutic logic and is directly testable: menin inhibitor plus serine/glycine-free medium in RS4;11.
  4. Is the residual 23% of PSAT1's fall that EPZ5676 does not reproduce a second input or noise? A dose-matched DOT1L arm would answer it, and the same experiment answers it for PHGDH, CBS and GPT2 at once.
Every source behind this entry (7)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset

    In RS4;11 at day 7, PSAT1 falls 2.40 log2 under VTP-50469 at 330 nM from a DMSO baseline of 8.35 log2 CPM and 1.85 log2 under EPZ5676 at 1 uM in the same experiment, a VTP-minus-EPZ margin of -0.55 with EPZ reproducing 77% of the effect - against PHGDH's margin of -1.72 and 47%; at day 2 PSAT1 is -1.05, 44% of its day-7 fall; in MV4;11 from an independent laboratory it is -2.17 under VTP-50469 500 nM against -1.19 under SGC0946 5 uM; and on the GPL570 array both RS4;11 probes fall under MI-2-2 (-0.77 at baseline 10.35, -0.49 at 11.67) while both REH probes rise (+1.18 at baseline 8.59, +0.49 at 10.65).

    What this evidence cannot show

    One line per platform, single doses, no dose-response, n=2-3 per arm, day 7 of continuous exposure so arrest is a competing explanation. MI-2-2 is the WEAK compound and moves PSAT1 only -0.77 in the fusion-positive line, so the REH contrast has little dynamic range and the wild-type null is correspondingly weak; GSE127506/GSE127508 contains no KMT2A-wild-type line, so the potent compound has never been run in one. Medium amino-acid composition is not recorded in any of the three series.

    Evidence that argues against this (1)
    • A 77%-reproduced effect is nearly compound-independent, so PSAT1 on its own would have supported the existing 'fusion-programme readout by any route' hypothesis rather than challenged it. The menin-selective reading of the serine arm rests on PHGDH, CBS and GPT2; if those three turn out to be dose artefacts the whole beat's verdict collapses and PSAT1 is one of the genes that would then look consistent. this repository 2026
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    PSAT1 DepMap gene effect is mean -0.2531, median -0.2431, with 8.2% of 1,258 screens below -0.5 and a leukaemia mean of -0.2669 over 57 screens, fetched with all 29 tissue blocks flattened and the resolved Ensembl id's approvedSymbol read back as PSAT1.

    What this evidence cannot show

    Rule 83: no release identifier in this payload, so depmap_release is null. The 57 leukaemia screens are matched on the payload's own diseaseFromSource string and are NOT split by KMT2A status. A knockout in serine-replete medium understates any dependency on a serine-synthesising enzyme. 8.2% is an absolute-threshold fraction, not a median split.

    Evidence that argues against this (1)
    • A preprint reports serine auxotrophy as a targetable vulnerability driven specifically by PSAT1 SUPPRESSION in AML - i.e. low PSAT1 creates the vulnerability rather than high PSAT1 creating the dependency. If that direction holds, a drug that lowers PSAT1 (as a menin inhibitor does here, -2.40) would CREATE a serine dependency rather than remove one, which inverts the therapeutic logic and is the more interesting prediction. Sinanidis I et al. 2025
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In AML, PSAT1 is upregulated in venetoclax-resistant cell lines and relapsed patient samples; the transcription factor ATF4 directly binds the PSAT1 promoter and enhances its expression, promoting glutathione synthesis, depleting the labile iron pool and attenuating lipid peroxidation; PSAT1 restrains JNK/c-Jun signalling; PSAT1 knockdown restores venetoclax sensitivity by triggering ferroptosis and modulating BCL-2 and GPX4; and elevated PSAT1 predicts poor survival.

    The ATF4/PSAT1/JNK signaling axis suppresses ferroptosis to drive venetoclax resistance in AML. · Zhu X, Huang W, Zhang M, et al., Zhang H · Exp Cell Res 2026
    PMID 42044713DOI
    What this evidence cannot show

    Abstract read this session; full text not retrieved, so the ChIP method, the promoter coordinates and the number of patient samples were not checked. The system is AML, not B-lineage KMT2A-rearranged ALL, and the fusion status of the lines is not stated in the abstract. 'Directly bound' in an abstract usually means ChIP or ChIP-qPCR plus a reporter; it is graded moderate here because the primary data were not seen. This is a 2026 paper in a non-specialist journal with no independent replication.

    Evidence that argues against this (1)
    • If ATF4 directly drives PSAT1 and ATF4 activity collapses under a menin inhibitor, then PSAT1's fall should be as compound-independent as CHAC1's - and it is 77% reproduced by the DOT1L inhibitor, close but not equal. The residual 23% is either a second input or noise, and this entry does not distinguish them. Meanwhile PHGDH, in the same pathway and the same cells, is only 47% reproduced, which no single-ATF4 model explains. this repository 2026
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Lycorine significantly inhibited growth and proliferation of the B-cell ALL lines REH and NALM-6 and induced apoptosis; RNA-seq identified PSAT1 of serine/glycine metabolism as its target in B-ALL cells; GSEA showed amino-acid metabolic pathway genes downregulated; ectopic expression, shRNA knockdown and LC-MS showed lycorine reduced serine/glycine metabolites by downregulating PSAT1; lycorine was synergistic with cytarabine; and it reduced leukaemia progression in a cell-line-derived xenograft.

    PMID 37951487DOI
    What this evidence cannot show

    Abstract read this session; full text not retrieved. **Both lines are KMT2A WILD TYPE** - REH is ETV6::RUNX1 and NALM-6 carries no KMT2A rearrangement - so this establishes PSAT1 as relevant in B-ALL generally and says nothing about the index fusion. Lycorine is a plant alkaloid with many reported targets; 'targeting PSAT1' here is an inference from transcriptomics plus rescue experiments, not a binding measurement, and a natural product with broad activity is a weak tool for a target claim. No infant-derived or primary sample.

    Evidence that argues against this (1)
    • This is the direct counter to reading 'PSAT1 is flat in REH under a menin inhibitor' as evidence that the wild-type line does not depend on PSAT1. It depends on it enough to be killed through it. The correct reading of the REH null is narrower: the KMT2A-fusion programme does not REGULATE PSAT1 in REH, which is a statement about regulation and not about requirement. this repository 2026
  5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProt reviewed human record Q9Y617 (SERC_HUMAN) has primary gene name PSAT1, the single synonym PSA, length 370 aa, catalyses the PLP-dependent transamination of 3-phosphohydroxypyruvate and L-glutamate to O-phosphoserine and alpha-ketoglutarate as the second step of phosphorylated L-serine biosynthesis, additionally acts as an inhibitor of ferroptosis in response to interferon-gamma, is expressed at high levels in brain, liver, kidney and pancreas and very weakly in thymus, prostate, testis and colon, carries three PDB cross-references, carries NO SUBCELLULAR LOCATION comment of any kind, and records two diseases - phosphoserine aminotransferase deficiency (low plasma and CSF serine and glycine, intractable seizures, acquired microcephaly, hypertonia, psychomotor retardation) and Neu-Laxova syndrome 2, a lethal autosomal recessive multiple-malformation syndrome.

    What this evidence cannot show

    Resolved 2026-09-22 with gene_exact:PSAT1 AND organism_id:9606 AND reviewed:true; exactly one reviewed row, gene_primary and length read back. The absent SUBCELLULAR LOCATION comment is a fact about the database, not the protein. The tissue-specificity statement is a curated summary of northern/EST-era data and should not be read as a quantitative expression profile. Disease descriptions are curated summaries, not cohorts.

    Evidence that argues against this (1)
    • The very weak thymus expression UniProt records is an awkward fact for a B-lineage argument: RS4;11 expresses PSAT1 at 8.35 log2 CPM and SEM at 12.07 log2 RMA on the array, well above background, so the leukaemic state is not the tissue UniProt describes and the curated profile is not a useful prior for this disease. this repository 2026
  6. E6established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Phosphoserine aminotransferase deficiency was described as a novel disorder of the serine biosynthesis pathway in 2007.

    Phosphoserine aminotransferase deficiency: a novel disorder of the serine biosynthesis pathway. · Hart CE, Race V, Achouri Y, et al., Van Schaftingen E · Am J Hum Genet 2007
    PMID 17436247DOI
    What this evidence cannot show

    Metadata verified against esummary on 2026-09-22; the ABSTRACT WAS NOT READ, so nothing beyond the title's assertion is claimed from it. The clinical description used in this entry comes from UniProt's curated DISEASE comment [E5], not from this paper. Cited here so that the founding clinical report is on the record rather than only the database summary - which is rule 91's point about a database row being a pointer.

    Evidence that argues against this (1)
    • A single founding case series cannot establish the phenotypic range of a disorder, and the later literature includes an adult diagnosis presenting as progressive neuropathy (PMID 34089226, metadata seen, not read), which is a far milder phenotype than the neonatal description. Whether pharmacological PSAT1 inhibition would look like the severe or the mild end is unknown and unknowable from these reports. Debs S et al. 2021
  7. E7established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    PubMed on 2026-09-22, one tag per term with translations read back: PSAT1[tiab] returns 352 records of which 1 is fusion-scoped (PMID 29672864) and 1 menin-scoped; PSAT1[tiab] AND leukemia[tiab] returns 8; PSAT[tiab] returns 327 with 0 fusion-scoped and EPIP[tiab] returns 62 with 0; and the UniProt-listed synonym PSA[tiab] returns 51,994 records, which are prostate-specific antigen and are excluded rather than unioned.

    What this evidence cannot show

    Bounded by titles and abstracts only; supporting figures not searched. The 8 leukaemia-scoped records were all opened by title and 5 are AML, 1 is B-ALL (PMID 37951487), 1 is a telomere prognostic model and 1 a DNMT3A/K562 GSH study. The 51,994 PSA records were not opened; the exclusion rests on the token's well-known referent rather than on an id-set intersection, which is a weaker basis than the SERA/PHGDH exclusion where zero overlap was measured.

    Evidence that argues against this (1)
    • Excluding PSA on prior knowledge rather than on a measured id-set intersection is exactly the shortcut this project's own rules warn against - the NMP22 case was a UniProt-listed synonym that LOOKED plausible and shared zero records. The right procedure would have been to intersect the 5 fusion-scoped PSA records with PSAT1's 1, and that was not done here. this repository 2026
Connected to