The serine/one-carbon arm is NOT a pure ATF4 output: it is menin-selective where the ISR arm is compound-independent (median margin -0.55 vs -0.02, p=0.011), and it is ABSENT in KMT2A::MLLT3 AML where the ISR arm is present

[
  "hyp-the-isr-collapse-is-a-fusion-programme-readout-not-a-menin-interface-readout",
  "hyp-menin-inhibition-collapses-the-serine-asns-programme-and-should-resensitise-kmt2a-r-all-to-asparaginase"
]

[
  "C:/Temp/oncofusion-geo-cache GSM1554911-GSM1554942 (GSE63664, GPL570 RMA; SEM, RS4;11, REH)",
  "C:/Temp/oncofusion-geo-cache/GPL570.txt (annotation; the multi-symbol /// field KEPT)",
  "C:/Temp/oncofusion-geo-cache/vtp_counts.txt.gz (GSE127506+GSE127508)",
  "C:/Temp/oncofusion-geo-cache/mv411_prc11.tab.gz (GSE260456)",
  "Open Targets GraphQL target.depMapEssentiality, all 29 tissue blocks flattened"
]

{
  "question": "Is the serine/one-carbon collapse under a menin inhibitor DOWNSTREAM of ATF4, or PARALLEL to it?",
  "answer": "NEITHER ALONE - BOTH INPUTS, and the ATF4 input is not sufficient. The pre-registered outcome reached is Outcome B: the serine arm is systematically less well reproduced by the DOT1L inhibitor than the ISR-output arm is, at margins comparable to the menin-target positive control.",
  "the_discriminating_number": "median VTP-minus-EPZ5676 margin: ISR output -0.024 (n=4), serine arm -0.55 (n=9), menin targets -1.018 (n=4); serine vs ISR Mann-Whitney p=0.01119, serine vs menin targets p=0.10629, ISR vs menin targets p=0.02857",
  "the_second_and_stronger_dissociation": "In MOLM13 (KMT2A::MLLT3 AML) the serine arm does not move at all (PHGDH -0.20 from a detected baseline of 7.22) while the ISR arm does (CHAC1 -1.89) and the positive control does (MEIS1 -2.96). The two arms therefore differ in fusion-partner/lineage dependence as well as in compound-selectivity.",
  "the_literature_half": "menin -> H3K4me3 at the PHGDH promoter -> de novo serine synthesis is PUBLISHED (PMID 29672864) and is ATF4-independent; menin -> ATF4 -> SSP transcripts and proteins is also published (PMID 33741715), including a one-sentence extension to MLL-rearranged B-ALL. Both, not either.",
  "what_would_refute_this": "a dose-response showing the serine margins collapse when EPZ5676 is titrated to match VTP-50469 on MEIS1"
}

{
  "PHGDH": -1.718,
  "CBS": -1.47,
  "GPT2": -1.031,
  "CTH": -0.752,
  "PSAT1": -0.552,
  "ALDH1L2": -0.489,
  "SHMT2": -0.423,
  "ASNS": -0.253,
  "MTHFD2": -0.032,
  "CHAC1": -0.085,
  "ATF5": -0.07,
  "TRIB3": -0.024,
  "DDIT3": 0.095,
  "MEIS1": -3.745,
  "MEF2C": -2.65,
  "PBX3": -1.018,
  "HOXA10": -0.568,
  "SLC3A2": -0.326,
  "SLC7A5": -0.193,
  "ATF4": 0.011,
  "housekeeping": {
    "ACTB": -0.032,
    "GAPDH": -0.2,
    "TBP": 0.061,
    "RPL13A": -0.166
  }
}

{
  "_all_baselines_above_the_1.0_log2cpm_floor": true,
  "positive_control_works": {
    "MEIS1": -2.956,
    "PBX3": -2.452,
    "MEF2C": -2.014
  },
  "serine_arm": {
    "PHGDH": -0.202,
    "PSAT1": -0.57,
    "PSPH": -0.306,
    "SHMT1": -0.476,
    "SHMT2": -0.408,
    "MTHFD1": 0.062,
    "MTHFD2": -0.292,
    "ALDH1L2": 0.077,
    "CBS": -1.021,
    "CTH": -0.368,
    "GPT2": -0.213,
    "GLS": -0.024,
    "ASNS": -0.406
  },
  "isr_arm_still_moves": {
    "CHAC1": -1.892,
    "DDIT3": -0.908,
    "ATF5": -0.737,
    "TRIB3": -0.457,
    "ATF4": -0.42
  },
  "baselines_log2cpm": {
    "PHGDH": 7.22,
    "PSAT1": 7.58,
    "ALDH1L2": 7.17,
    "MTHFD2": 8.09,
    "CBS": 5.62,
    "CHAC1": 5.59
  },
  "rs411_versus_molm13_difference_for_phgdh_log2": 3.04,
  "reading": "The only gene in the serine arm to move past 1 log2 in MOLM13 is CBS (-1.021), and CBS is also the gene with the largest platform disagreement in this run. The serine arm as measured here is a KMT2A::AFF1 / B-lineage phenomenon."
}

{
  "serine_median_menin_minus_dot1l": -1.051,
  "serine_n": 7,
  "isr_median_menin_minus_dot1l": 0.016,
  "isr_n": 2,
  "largest": {
    "PHGDH": -1.889,
    "ALDH1L2": -1.889,
    "CBS": -1.053,
    "GPT2": -1.05,
    "PSAT1": -0.977
  },
  "caveat": "different DOT1L compound (SGC0946 5 uM) at a different dose, and housekeeping is not flat here (ACTB -0.42, TBP -0.47, RPL13A -0.73 margins), so part of the -1.05 median is a normalisation artefact. What survives the correction is the CONTRAST with the ISR arm's +0.016, and the ISR arm rests on only 2 usable genes because DDIT3, CHAC1 and TRIB3 are near the floor in MV4;11."
}

{
  "serine_median": 0.384,
  "isr_output_median": 0.239,
  "serine": {
    "GPT2": 0.58,
    "ASNS": 0.48,
    "CBS": 0.46,
    "PSAT1": 0.44,
    "ALDH1L2": 0.38,
    "PHGDH": 0.33,
    "MTHFD2": 0.24,
    "SHMT2": 0.2,
    "CTH": -0.07
  },
  "isr_output": {
    "CHAC1": 0.28,
    "DDIT3": 0.24,
    "ATF5": 0.22,
    "TRIB3": 0.13
  },
  "menin_targets": {
    "MEF2C": 0.76,
    "PBX3": 0.64,
    "HOXA10": 0.6,
    "MEIS1": 0.49
  },
  "reading": "The serine arm is earlier than the ISR-output arm and the menin targets are earlier still, consistent with the ordering menin target -> serine -> ISR output, which is the ordering a two-input model with a chromatin-proximal serine input predicts. It is NOT proof: a 37% median difference at two coarse timepoints cannot separate a direct effect from a faster indirect one, and CTH's negative share (+0.10 at day 2 against -1.57 at day 7) shows how unstable these ratios are at small day-2 values."
}

{
  "direction": "every detected serine gene is FLAT or POSITIVE in REH under MI-2-2 while falling in RS4;11 and SEM - PHGDH +0.34/+0.21, PSAT1 +1.18/+0.49, SHMT2 +0.17/+0.04/+0.15, MTHFD2 +0.41/+0.70, ASNS +0.26, CTH +0.43/+0.47",
  "why_the_null_is_WEAK_and_must_be_labelled": "the wild-type arm exists only for MI-2-2, the compound that moves PHGDH just -0.44 and PSAT1 -0.77 in the fusion-positive lines, so the contrast is against tenths of a log2 and not against the -3.24 the potent compound produces. GSE127506/GSE127508 contains NO KMT2A-wild-type line at all.",
  "GPT2_cannot_be_tested_in_REH": "REH baseline 5.17 log2 RMA, below the 6.0 floor, against 10.78 in RS4;11 - a new instance of the STC2/SOCS2 problem the cycle-58 beat found, on a different gene",
  "and_REH_is_killed_through_PSAT1_anyway": "PMID 37951487: lycorine eliminates REH and NALM-6 by downregulating PSAT1, with cytarabine synergy and xenograft activity. A flat regulation response in REH is not evidence that REH does not require the pathway."
}

{
  "floor_collisions_detected": 8,
  "floor_collision_members": "HOXA9 only - RS4;11 d2 (baseline 0.00, both arms), RS4;11 d7 (baseline 0.763, lfc -0.763 in BOTH arms), MOLM13 d2 and d7 (baseline 0.00). Excluded from the menin-target positive control for that reason; the control rests on MEIS1, MEF2C, PBX3 and HOXA10.",
  "tyms_baseline_zero": "TYMS RS4;11 d7 DMSO baseline is 0.00 log2 CPM; its +0.81 is not a measurement. TYMS is a declared context gene and carries no part of the verdict.",
  "cbs_platform_disagreement": "CBS is below the array floor on all three GPL570 probes in RS4;11 (4.83, 4.07, 3.86) and well expressed in SEM (8.68, 9.78) - and on RNA-seq the RS4;11 baseline is 5.97 log2 CPM, detected. So the gene with the largest RNA-seq fall in the panel (-4.07) is a gene the array says RS4;11 barely expresses. Not resolvable from these data; recorded, not explained.",
  "aldh1l2_array_coverage": "only one of three GPL570 probes (231202_at) is above floor in RS4;11 (6.47); the same probe is at 3.02 in SEM and 3.14 in REH. So ALDH1L2 has no usable wild-type array control at all.",
  "cth_mv411_non_replication_is_a_floor_artefact": "the cycle-56 entry lists CTH under 'did_not_replicate_the_isr_output' as '+0.24 against -1.57 in RS4;11'. CTH's MV4;11 Safe3 DMSO baseline is 0.29 log2 CPM, BELOW the 1.0 floor. That is an absence of signal, not a failure to replicate, and the entry should say so. Reported, not edited - the entry is not owned by this beat.",
  "slc7a11_is_the_one_dot1l_SELECTIVE_gene_in_the_panel": "RS4;11 d7: VTP-50469 +0.13, EPZ5676 -1.84, margin +1.96, from a detected baseline of 7.04 log2 CPM. It is the only gene in 41 whose margin is positive and large. On the array both compounds lower it in RS4;11 and SEM. Unexplained and recorded as an anomaly rather than a finding; SLC7A11 is the cystine/glutamate antiporter and sits downstream of the transsulfuration arm, so it is not a random gene to behave oddly here.",
  "molm13_mthfd2l_rises": "MTHFD2L +0.97 VTP / +1.45 EPZ in MOLM13 from a baseline of 1.05 log2 CPM - at the floor, so not interpreted"
}

{
  "GLS": "NOT part of the programme. RS4;11 d7 VTP -0.06, EPZ +0.31, baseline 6.84 detected; MOLM13 -0.02; MV4;11 -0.54; and on the array all eight GPL570 probes are within 0.49 in every line. DepMap mean -0.1803, 10.1% below -0.5. No entry written: a glutaminase that does not move under the drug is not part of this beat's biology, whatever its standing in cancer metabolism generally. PubMed: GLS[tiab] returns 6,986 records and GLS[tiab] AND the fusion operand set returns 0 with a non-empty translation and warninglist 'No items found.'",
  "SHMT1": "moves in the WRONG DIRECTION: RS4;11 d7 VTP +0.78, EPZ +1.05, from a baseline of 3.02 log2 CPM. The cytosolic paralogue rises while the mitochondrial SHMT2 falls (-1.09). On the array all four probes are within 0.37 in every line. Recorded as a paralogue-direction observation inside gene-shmt2 rather than given an entry of its own, because the +0.78 sits on a low baseline and is not independently interesting.",
  "MTHFD2L": "baseline 2.23 log2 CPM in RS4;11 and 1.05 in MOLM13, at or near the floor, and all seven array probes flat. Not measurable here. DepMap mean -0.0721.",
  "MTR_MTHFR_DHFR_TYMS": "declared in the pre-registration as CONTEXT genes and not part of the verdict. All flat in RS4;11 (MTR +0.46, MTHFR -0.05, DHFR +0.29) or at the floor (TYMS baseline 0.00). Their DepMap profiles are the interesting part and are recorded in the DepMap results file: DHFR -0.8234 (69.8%), TYMS -0.7992 (51.2%), MAT2A -1.3510 (95.2%) - the one-carbon SINKS are pan-essential while the serine SOURCES are not.",
  "SLC1A5_SLC7A11": "|VTP| below 1 log2 at RS4;11 day 7 (-0.99 and +0.13), so excluded from the transport arm by the pre-declared rule, leaving that arm with only SLC3A2 and SLC7A5 and too small to compare. No transporter entry written."
}

{
  "payload_guard": "panel maximum flattened screens = 1,258, threshold 1,000, PASS; all 29 tissue blocks flattened per gene and n_tissue_blocks recorded so a [0]-only read would be detectable",
  "readback": "every resolved Ensembl id's approvedSymbol compared back to the symbol asked for; 41 of 41 matched",
  "depmap_release": null,
  "the_inverse_relationship": "the genes that MOVE are not required and the gene that is required does not move. PHGDH -0.0795 (1.1% of 1,258 below -0.5), ALDH1L2 +0.0065 (0.0%), CBS -0.0256 (0.4%), CTH +0.0699 (0.0%), GPT2 -0.0882 (0.5%), PSPH -0.1161 (0.6%), MTHFD2 -0.0282 (1.7%), PSAT1 -0.2531 (8.2%), ASNS -0.2973 (17.5%) - against MTHFD1 -0.3512 (28.8%, leukaemia -0.744), which is flat under both drugs (+0.09 VTP, +0.36 EPZ).",
  "a_guard_that_did_NOT_fire": "POLR2A, included as a pan-essential reference control, returned 0 tissue blocks and 0 screens - an EMPTY payload for the canonical pan-essential gene, with the readback passing. The panel-maximum guard cannot see a single-gene empty payload, only a truncated one. Recorded as a refinement to rule 5: a panel-maximum guard needs a companion per-gene zero-screen check, which this script prints but did not enforce."
}