{
  "acceptance_version": "0.1",
  "object_path": "examples/quality-control-for-modern-bone-collagen-stable-carbon-and-n",
  "manifest": "agent/tools.json",
  "published_results_path": "results.json",
  "notes": "Every published unit is reproduced at the tool defaults and cross-linked to its recipe_id; the four data-subset robustness variants are additionally reproduced through fish_vs_mammal_contrast's own arguments. Off-reference literals are tool outputs, except the two Table S2 calls, whose literals are the source's own printed statistics. Several qualify published claims or their framing. The nonparametric justification rests on two samples (normality-ceiling-table-s2). Claim 5's 'not a pass/fail effect' holds only for the 2.9-3.6 window (window-3.1-3.5-off-reference). Claim 3's glycine-first ranking does not hold within mammals (t3-within-mammals-off-reference), and its imino-acid ranking is partly a between-class effect (t3-within-fish-off-reference). Claim 1's CI belongs to the all-tissue contrast, not bone collagen alone (t1-bone-collapsed-off-reference). The Material code treated as a batch factor is very probably tissue (normality-bone-stratum-table-s2). None overturns claims 1, 2 or 4.",
  "calls": [
    {
      "call_id": "integrity-published",
      "tool": "data_integrity",
      "args": {},
      "expect": {"kind": "published_pointer", "result_path": "/data_integrity"},
      "recipe_id": "bone-collagen-qc-integrity",
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true,
      "notes": "Takes no parameters. The five tables are fetched through scripts/fetch_pinned.py and then re-hashed by the analysis unit's own acquire_and_verify."
    },
    {
      "call_id": "dataset-published",
      "tool": "dataset_summary",
      "args": {},
      "expect": {"kind": "published_pointer", "result_path": "/dataset"},
      "recipe_id": "bone-collagen-qc-dataset",
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true
    },
    {
      "call_id": "normality-published",
      "tool": "normality_check",
      "args": {},
      "expect": {"kind": "published_pointer", "result_path": "/normality_shapiro"},
      "recipe_id": "bone-collagen-qc-normality",
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true
    },
    {
      "call_id": "normality-ceiling-table-s2",
      "tool": "normality_check",
      "args": {"max_cn": 3.5},
      "expect": {"kind": "literal", "value": {"cn_fish": {"n": 288, "W": 0.9949, "p": 0.4585}, "cn_mammal": {"n": 86, "W": 0.9764, "p": 0.1162}}},
      "rel_tolerance": 0.0,
      "abs_tolerance": 5e-05,
      "enabled": true,
      "notes": "The literal is not a tool output: it is the source's own Table S2 'All' rows for Actinopterygii and Mammalia, read from the pinned file at their printed 4 decimal places, so the tolerance is half the last place. Dropping the two samples above C:N 3.5 reproduces both rows. The object's stated justification for going nonparametric ('non-normal by Shapiro-Wilk') therefore rests on two fish from one 1968 study; the mammal side was never non-normal (p = 0.12). Nonparametric tests remain a defensible choice, but not for the reason given."
    },
    {
      "call_id": "normality-leave-out-rigby-1968",
      "tool": "normality_check",
      "args": {"exclude_references": ["Rigby 1968"]},
      "expect": {
        "kind": "literal",
        "value": {"cn_fish": {"n": 283, "W": 0.9951194380889378, "p": 0.5117894501332412}, "cn_mammal": {"n": 86, "W": 0.9763787528626732, "p": 0.1162177911487477}}
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true,
      "notes": "The study-level version of the same check: leaving out Rigby 1968 (7 fish, including both samples above 3.5) makes fish C:N consistent with normality, W = 0.995, p = 0.51 on n = 283."
    },
    {
      "call_id": "normality-bone-stratum-table-s2",
      "tool": "normality_check",
      "args": {"fish_classes": ["Elasmobranchii"], "material": 1},
      "expect": {"kind": "literal", "value": {"cn_fish": {"n": 6, "W": 0.9508, "p": 0.7466}, "cn_mammal": {"n": 24, "W": 0.9534, "p": 0.3202}}},
      "rel_tolerance": 0.0,
      "abs_tolerance": 5e-05,
      "enabled": true,
      "notes": "The literal is the source's Table S2 Bone rows for Elasmobranchii (n = 6) and Mammalia (n = 24), read from the pinned file at 4 decimal places. Material = 1 reproduces both exactly, and Material = 2 holds 50 ray-finned fish and nothing else, as a scale stratum would. That is strong evidence that the Material code the object calls an undocumented batch factor is tissue, with 1 = bone. It is not proof: the ray-finned fish counts differ from Table S2's tissue rows by 1 to 3 in every stratum (Material 1/2/3: 51/50/189 against Bone/Scale/Skin 50/49/186), and mammal Material = 3 by one (62 against Skin 61). If the mapping holds, R1 is a within-tissue stratification, and its Material = 1 stratum is the only bone-collagen contrast built on a dataset titled for bone collagen."
    },
    {
      "call_id": "t1-published",
      "tool": "fish_vs_mammal_contrast",
      "args": {},
      "expect": {"kind": "published_pointer", "result_path": "/T1_fish_vs_mammal_CN"},
      "recipe_id": "bone-collagen-qc-t1-cn",
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true
    },
    {
      "call_id": "r1-material-1-published",
      "tool": "fish_vs_mammal_contrast",
      "args": {"material": 1},
      "select": "/p",
      "expect": {"kind": "published_pointer", "result_path": "/robustness/variants/1/p"},
      "recipe_id": "bone-collagen-qc-robustness",
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true
    },
    {
      "call_id": "r1-material-3-published",
      "tool": "fish_vs_mammal_contrast",
      "args": {"material": 3},
      "select": "/p",
      "expect": {"kind": "published_pointer", "result_path": "/robustness/variants/3/p"},
      "recipe_id": "bone-collagen-qc-robustness",
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true
    },
    {
      "call_id": "r2-collapse-published",
      "tool": "fish_vs_mammal_contrast",
      "args": {"collapse_taxa": true},
      "select": "/p",
      "expect": {"kind": "published_pointer", "result_path": "/robustness/variants/4/p"},
      "recipe_id": "bone-collagen-qc-robustness",
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true
    },
    {
      "call_id": "r2-collapse-ci-published",
      "tool": "fish_vs_mammal_contrast",
      "args": {"collapse_taxa": true},
      "select": "/hl_ci95/0",
      "expect": {"kind": "published_pointer", "result_path": "/robustness/variants/4/ci_lo"},
      "recipe_id": "bone-collagen-qc-robustness",
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true,
      "notes": "The lower CI bound of R2 is the one value that depends on the collapsed rows' order, because the bootstrap draws indices. Matching it shows the collapse preserves main()'s ordering."
    },
    {
      "call_id": "r3-broadened-published",
      "tool": "fish_vs_mammal_contrast",
      "args": {"fish_classes": ["Actinopterygii", "Elasmobranchii"], "mammal_classes": ["Mammalia", "Aves"]},
      "select": "/p",
      "expect": {"kind": "published_pointer", "result_path": "/robustness/variants/5/p"},
      "recipe_id": "bone-collagen-qc-robustness",
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true
    },
    {
      "call_id": "t1-bone-collapsed-off-reference",
      "tool": "fish_vs_mammal_contrast",
      "args": {"material": 1, "collapse_taxa": true},
      "expect": {
        "kind": "literal",
        "value": {
          "test": "Mann-Whitney U (two-sided)",
          "U": 78.0,
          "p": 0.0027159716815677295,
          "cliffs_delta": -0.59375,
          "hodges_lehmann_shift_fish_minus_mammal": -0.04499999999999993,
          "hl_ci95": [-0.07000000000000028, -0.020000000000000018],
          "group_fish_actinopterygii": {"n": 32, "median": 3.19, "iqr_lo": 3.15, "iqr_hi": 3.2125, "mean": 3.1834375, "sd": 0.041879964799566176, "min": 3.1, "max": 3.26, "pct_in_archaeo_window": 100.0},
          "group_mammal_mammalia": {"n": 12, "median": 3.22, "iqr_lo": 3.21, "iqr_hi": 3.25125, "mean": 3.2320833333333336, "sd": 0.04008277420387444, "min": 3.18, "max": 3.32, "pct_in_archaeo_window": 100.0},
          "expected_direction": "fish < mammal",
          "observed_direction": "fish < mammal"
        }
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true,
      "notes": "The published battery applies the batch stratum (R1) and the per-species collapse (R2) separately, never together. Together they are the strictest combination of the battery's own guards: Material = 1 (bone, see normality-bone-stratum-table-s2) with one median row per named taxon, leaving 32 fish rows (31 taxa plus one unnamed sample kept individually) against 12 mammal taxa. The contrast survives (p = 0.0027, Cliff's delta -0.59) with a smaller shift than the published bone stratum (-0.045 against R1's -0.050) and a wider CI (-0.07 to -0.02 against R1's -0.07 to -0.03); that interval is two and a half times the width of claim 1's pooled-tissue CI (-0.070 to -0.050). Claim 4 holds. The precision in claim 1 belongs to the all-tissue contrast; a reading restricted to bone collagen should carry the wider interval, which the published R1 stratum already shows in part."
    },
    {
      "call_id": "t1-leave-out-rigby-1968-off-reference",
      "tool": "fish_vs_mammal_contrast",
      "args": {"exclude_references": ["Rigby 1968"]},
      "expect": {
        "kind": "literal",
        "value": {
          "test": "Mann-Whitney U (two-sided)",
          "U": 4717.0,
          "p": 6.904797107062866e-18,
          "cliffs_delta": -0.6123757087681814,
          "hodges_lehmann_shift_fish_minus_mammal": -0.06000000000000005,
          "hl_ci95": [-0.07000000000000028, -0.04999999999999982],
          "group_fish_actinopterygii": {"n": 283, "median": 3.16, "iqr_lo": 3.12, "iqr_hi": 3.2, "mean": 3.157349823321555, "sd": 0.05415083895839246, "min": 3.0, "max": 3.32, "pct_in_archaeo_window": 100.0},
          "group_mammal_mammalia": {"n": 86, "median": 3.22, "iqr_lo": 3.1825, "iqr_hi": 3.24, "mean": 3.2174418604651183, "sd": 0.04467361496616324, "min": 3.11, "max": 3.33, "pct_in_archaeo_window": 100.0},
          "expected_direction": "fish < mammal",
          "observed_direction": "fish < mammal"
        }
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true,
      "notes": "Leave-one-study-out on the study that supplies both outlying fish. The contrast is unchanged in shift (-0.060) and slightly stronger (p = 6.9e-18, delta -0.61): the headline does not lean on the samples that drive the normality result."
    },
    {
      "call_id": "t1-elasmobranchs-off-reference",
      "tool": "fish_vs_mammal_contrast",
      "args": {"fish_classes": ["Elasmobranchii"]},
      "expect": {
        "kind": "literal",
        "value": {
          "test": "Mann-Whitney U (two-sided)",
          "U": 826.5,
          "p": 0.0017191857571564333,
          "cliffs_delta": -0.3799699924981246,
          "hodges_lehmann_shift_fish_minus_mammal": -0.03000000000000025,
          "hl_ci95": [-0.050000000000000266, -0.010000000000000231],
          "group_fish_elasmobranchii": {"n": 31, "median": 3.18, "iqr_lo": 3.1550000000000002, "iqr_hi": 3.225, "mean": 3.1832258064516132, "sd": 0.04791569657737039, "min": 3.08, "max": 3.28, "pct_in_archaeo_window": 100.0},
          "group_mammal_mammalia": {"n": 86, "median": 3.22, "iqr_lo": 3.1825, "iqr_hi": 3.24, "mean": 3.2174418604651183, "sd": 0.04467361496616324, "min": 3.11, "max": 3.33, "pct_in_archaeo_window": 100.0},
          "expected_direction": "fish < mammal",
          "observed_direction": "fish < mammal"
        }
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true,
      "notes": "Cartilaginous fish alone against mammals: the offset is present (p = 0.0017) but half the size, HL -0.030 (CI -0.050 to -0.010), Cliff's delta -0.38 against -0.60 for ray-finned fish. R3 pools elasmobranchs with ray-finned fish as 'fish'; they are not interchangeable, and the broadened estimate is carried by the 290 Actinopterygii."
    },
    {
      "call_id": "t1-fish-vs-birds-off-reference",
      "tool": "fish_vs_mammal_contrast",
      "args": {"mammal_classes": ["Aves"]},
      "expect": {
        "kind": "literal",
        "value": {
          "test": "Mann-Whitney U (two-sided)",
          "U": 457.5,
          "p": 0.0008932224692894528,
          "cliffs_delta": -0.6494252873563218,
          "hodges_lehmann_shift_fish_minus_mammal": -0.06000000000000005,
          "hl_ci95": [-0.060124999999999935, -0.040000000000000036],
          "group_fish_actinopterygii": {"n": 290, "median": 3.16, "iqr_lo": 3.12, "iqr_hi": 3.2, "mean": 3.160137931034483, "sd": 0.06369245099730167, "min": 3.0, "max": 3.58, "pct_in_archaeo_window": 100.0},
          "group_mammal_aves": {"n": 9, "median": 3.22, "iqr_lo": 3.21, "iqr_hi": 3.22, "mean": 3.21, "sd": 0.016583123951777082, "min": 3.17, "max": 3.22, "pct_in_archaeo_window": 100.0},
          "expected_direction": "fish < mammal",
          "observed_direction": "fish < mammal"
        }
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true,
      "notes": "Ray-finned fish against birds alone (n = 9): the same -0.060 shift as against mammals (p = 0.00089). R3's pooling of birds with mammals is supported: birds sit with mammals, not with fish."
    },
    {
      "call_id": "t1-bootstrap-stability-off-reference",
      "tool": "fish_vs_mammal_contrast",
      "args": {"seed": 1, "n_boot": 20000},
      "select": "/hl_ci95",
      "expect": {"kind": "literal", "value": [-0.06999999999999984, -0.04999999999999982]},
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true,
      "notes": "A different seed and twice the resamples return the identical interval, -0.07 to -0.05. C:N is reported to two decimals, so pairwise differences sit on a 0.01 grid and the percentile bounds are set by the data, not by Monte Carlo noise."
    },
    {
      "call_id": "robustness-published",
      "tool": "robustness_battery",
      "args": {},
      "expect": {"kind": "published_pointer", "result_path": "/robustness"},
      "recipe_id": "bone-collagen-qc-robustness",
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true,
      "notes": "Takes no parameters: it is the fixed pre-registered battery. Every variant in it is a fish_vs_mammal_contrast call."
    },
    {
      "call_id": "qc-window-fish-published",
      "tool": "qc_window_pass_rate",
      "args": {},
      "select": "/fish/pct_in_window",
      "expect": {"kind": "published_pointer", "result_path": "/T1_fish_vs_mammal_CN/group_fish_actinopterygii/pct_in_archaeo_window"},
      "recipe_id": "bone-collagen-qc-t1-cn",
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true
    },
    {
      "call_id": "qc-window-mammal-published",
      "tool": "qc_window_pass_rate",
      "args": {},
      "select": "/mammal/pct_in_window",
      "expect": {"kind": "published_pointer", "result_path": "/T1_fish_vs_mammal_CN/group_mammal_mammalia/pct_in_archaeo_window"},
      "recipe_id": "bone-collagen-qc-t1-cn",
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true
    },
    {
      "call_id": "qc-window-all-samples",
      "tool": "qc_window_pass_rate",
      "args": {},
      "select": "/all_samples",
      "expect": {"kind": "literal", "value": {"n": 436, "n_in_window": 436, "pct_in_window": 100.0}},
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true,
      "notes": "At the published window, not off it. results.json reports the pass rate only for the two focal classes, but claim 5's subject reads 'all 436 modern collagen samples'. This checks that wording: all 436, across all ten classes, lie in 2.9-3.6."
    },
    {
      "call_id": "window-3.1-3.5-off-reference",
      "tool": "qc_window_pass_rate",
      "args": {"window_lo": 3.1, "window_hi": 3.5},
      "expect": {
        "kind": "literal",
        "value": {
          "window": [3.1, 3.5],
          "fish": {"n": 290, "n_in_window": 256, "pct_in_window": 88.27586206896552},
          "mammal": {"n": 86, "n_in_window": 86, "pct_in_window": 100.0},
          "all_samples": {"n": 436, "n_in_window": 395, "pct_in_window": 90.59633027522935}
        }
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true,
      "notes": "Claim 5 says the class difference is 'a systematic offset, not a pass/fail effect'. Under the stricter 3.1-3.5 window it becomes one: 34 of 290 ray-finned fish fail (32 below 3.1, drawn from 26 different studies, plus the two Rigby 1968 samples above 3.5) and all 86 mammals pass. Claim 5 names the 2.9-3.6 window for its 100% pass rates, but its closing 'not a pass/fail effect' is true of that window only and should be scoped to it; a laboratory applying the tighter window would reject fish collagen preferentially."
    },
    {
      "call_id": "t2-published",
      "tool": "glycine_association",
      "args": {},
      "expect": {"kind": "published_pointer", "result_path": "/T2_CN_vs_glycine"},
      "recipe_id": "bone-collagen-qc-t2-glycine",
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true
    },
    {
      "call_id": "t2-elasmobranchs-off-reference",
      "tool": "glycine_association",
      "args": {"within_classes": ["Elasmobranchii"]},
      "expect": {
        "kind": "literal",
        "value": {
          "test": "Spearman rank correlation (C:N vs glycine, residues/1000)",
          "overall": {"rho": -0.6160052561250701, "p": 6.643748814276406e-47, "n": 436},
          "elasmobranchii": {"rho": -0.6832910990952478, "p": 2.2722950453926195e-05, "n": 31},
          "expected_direction": "negative"
        }
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true,
      "notes": "An out-of-sample replication: the confirmatory test used only ray-finned fish and mammals, and elasmobranchs (n = 31) show the same negative association at least as strongly (rho = -0.68, p = 2.3e-05, against -0.63 in ray-finned fish and -0.47 in mammals)."
    },
    {
      "call_id": "t2-taxon-collapsed-off-reference",
      "tool": "glycine_association",
      "args": {"collapse_taxa": true},
      "expect": {
        "kind": "literal",
        "value": {
          "test": "Spearman rank correlation (C:N vs glycine, residues/1000)",
          "overall": {"rho": -0.6539691029698179, "p": 3.7538345004176373e-26, "n": 203},
          "actinopterygii": {"rho": -0.6582094006781152, "p": 5.543529530012908e-19, "n": 142},
          "mammalia": {"rho": -0.5766637990597792, "p": 0.003971200086406675, "n": 23},
          "expected_direction": "negative"
        }
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true,
      "notes": "The pseudoreplication guard (R2) was applied only to T1. Applied to T2 it leaves 203 rows (197 per-taxon medians plus 6 unnamed fish samples kept individually), and the association strengthens overall (rho = -0.65) and survives within mammals on 23 taxa (rho = -0.58, p = 0.004)."
    },
    {
      "call_id": "t2-bone-stratum-off-reference",
      "tool": "glycine_association",
      "args": {"material": 1},
      "expect": {
        "kind": "literal",
        "value": {
          "test": "Spearman rank correlation (C:N vs glycine, residues/1000)",
          "overall": {"rho": -0.6459007002463322, "p": 1.0823071497778873e-11, "n": 88},
          "actinopterygii": {"rho": -0.6353362654213657, "p": 5.489793208682646e-07, "n": 51},
          "mammalia": {"rho": -0.8666729564285887, "p": 4.344835547086734e-08, "n": 24},
          "expected_direction": "negative"
        }
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true,
      "notes": "Within the Material = 1 (bone) stratum the association holds in both classes, and within mammal bone it is far stronger than the published mammal estimate (rho = -0.87 on n = 24, against -0.47 on all 86 mammals)."
    },
    {
      "call_id": "t3-published",
      "tool": "amino_acid_screen",
      "args": {},
      "expect": {"kind": "published_pointer", "result_path": "/T3_aa_screen"},
      "recipe_id": "bone-collagen-qc-t3-aa-screen",
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true
    },
    {
      "call_id": "t3-within-mammals-off-reference",
      "tool": "amino_acid_screen",
      "args": {"classes": ["Mammalia"]},
      "select": "/results",
      "expect": {
        "kind": "literal",
        "value": [
          {"residue": "Ala", "residue_full": "Alanine", "rho": -0.4860910943114541, "p": 2.0886871440640013e-06, "n": 86, "q_bh": 1.3228351912405342e-05, "significant_q<0.05": true},
          {"residue": "Gly", "residue_full": "Glycine", "rho": -0.47454711624181256, "p": 3.9117680774729405e-06, "n": 86, "q_bh": 1.8580898367996467e-05, "significant_q<0.05": true},
          {"residue": "Arg", "residue_full": "Arginine", "rho": -0.2770287537215793, "p": 0.009816820576846068, "n": 86, "q_bh": 0.02331494887000941, "significant_q<0.05": true},
          {"residue": "Hyl", "residue_full": "Hydroxylysine", "rho": -0.1670306597797633, "p": 0.16384795656979706, "n": 71, "q_bh": 0.22236508391615314, "significant_q<0.05": false},
          {"residue": "Met", "residue_full": "Methionine", "rho": -0.0934052601267962, "p": 0.3923287789241671, "n": 86, "q_bh": 0.4384851058564221, "significant_q<0.05": false},
          {"residue": "His", "residue_full": "Histidine", "rho": 0.06277275842834967, "p": 0.5658466814720586, "n": 86, "q_bh": 0.5972826082205063, "significant_q<0.05": false},
          {"residue": "Cys", "residue_full": "Cysteine", "rho": 0.07111181345347778, "p": 0.7862284958774103, "n": 17, "q_bh": 0.7862284958774103, "significant_q<0.05": false},
          {"residue": "Glu", "residue_full": "Glutamate", "rho": 0.12154337773852325, "p": 0.2649394580094755, "n": 86, "q_bh": 0.31461560638625213, "significant_q<0.05": false},
          {"residue": "Lys", "residue_full": "Lysine", "rho": 0.12801406183981928, "p": 0.2401459789105199, "n": 86, "q_bh": 0.30418490661999187, "significant_q<0.05": false},
          {"residue": "Val", "residue_full": "Valine", "rho": 0.17549959599382467, "p": 0.10603552005879013, "n": 86, "q_bh": 0.15497499085515481, "significant_q<0.05": false},
          {"residue": "Thr", "residue_full": "Threonine", "rho": 0.17754332307416404, "p": 0.10196586258164807, "n": 86, "q_bh": 0.15497499085515481, "significant_q<0.05": false},
          {"residue": "Ser", "residue_full": "Serine", "rho": 0.19448619134241205, "p": 0.07275340895543471, "n": 86, "q_bh": 0.12566497910484176, "significant_q<0.05": false},
          {"residue": "Hyp", "residue_full": "Hydroxyproline", "rho": 0.20197898762567706, "p": 0.06219013174245932, "n": 86, "q_bh": 0.1181612503106727, "significant_q<0.05": false},
          {"residue": "Pro", "residue_full": "Proline", "rho": 0.2157626485954095, "p": 0.046022332162852495, "n": 86, "q_bh": 0.09715825678824415, "significant_q<0.05": false},
          {"residue": "Tyr", "residue_full": "Tyrosine", "rho": 0.3033011183162282, "p": 0.004529273437019466, "n": 86, "q_bh": 0.012293742186195694, "significant_q<0.05": true},
          {"residue": "Ile", "residue_full": "Isoleucine", "rho": 0.38810854997321104, "p": 0.00022219430644925897, "n": 86, "q_bh": 0.0007036153037559868, "significant_q<0.05": true},
          {"residue": "Asp", "residue_full": "Aspartate", "rho": 0.41368907109541825, "p": 7.521584698545568e-05, "n": 86, "q_bh": 0.00028582021854473157, "significant_q<0.05": true},
          {"residue": "Leu", "residue_full": "Leucine", "rho": 0.489373325567664, "p": 1.7400662432732333e-06, "n": 86, "q_bh": 1.3228351912405342e-05, "significant_q<0.05": true},
          {"residue": "Phe", "residue_full": "Phenylalanine", "rho": 0.6388070763309555, "p": 3.641738769807303e-11, "n": 86, "q_bh": 6.919303662633876e-10, "significant_q<0.05": true}
        ]
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true,
      "notes": "Claim 3 says glycine is the strongest negative correlate of C:N. Within mammals it is not: alanine edges it (rho = -0.486 against -0.475), proline and hydroxyproline are not FDR-significant (q = 0.097 and 0.118), and the strongest positive correlate is phenylalanine (0.64), not leucine. Glycine's first place in the pooled screen holds within ray-finned fish (rho = -0.63, t3-within-fish-off-reference), which supply 290 of the 436 rows, but not within mammals. The pooled positive ranking matches neither class (see t3-within-fish-off-reference for fish). Claim 3, already labelled exploratory, should say its ranking is pooled across classes."
    },
    {
      "call_id": "t3-within-fish-off-reference",
      "tool": "amino_acid_screen",
      "args": {"classes": ["Actinopterygii"]},
      "select": "/results",
      "expect": {
        "kind": "literal",
        "value": [
          {"residue": "Gly", "residue_full": "Glycine", "rho": -0.6302391831522303, "p": 1.640273992019731e-33, "n": 290, "q_bh": 3.1165205848374887e-32, "significant_q<0.05": true},
          {"residue": "Ser", "residue_full": "Serine", "rho": -0.26413452328235637, "p": 5.118789218938721e-06, "n": 290, "q_bh": 1.08063327955373e-05, "significant_q<0.05": true},
          {"residue": "Arg", "residue_full": "Arginine", "rho": -0.212100498673429, "p": 0.00027495858836411134, "n": 290, "q_bh": 0.0004749284708107378, "significant_q<0.05": true},
          {"residue": "His", "residue_full": "Histidine", "rho": -0.1471585036846417, "p": 0.012111830340020713, "n": 290, "q_bh": 0.017701905881568735, "significant_q<0.05": true},
          {"residue": "Ala", "residue_full": "Alanine", "rho": -0.13363254526717852, "p": 0.02284185187040227, "n": 290, "q_bh": 0.030999656109831655, "significant_q<0.05": true},
          {"residue": "Hyl", "residue_full": "Hydroxylysine", "rho": -0.05979812147201824, "p": 0.37413552956648244, "n": 223, "q_bh": 0.41815147422136273, "significant_q<0.05": false},
          {"residue": "Cys", "residue_full": "Cysteine", "rho": -0.031489194694875225, "p": 0.7748103761465864, "n": 85, "q_bh": 0.7748103761465864, "significant_q<0.05": false},
          {"residue": "Met", "residue_full": "Methionine", "rho": 0.023352121348880282, "p": 0.6926071211807903, "n": 289, "q_bh": 0.731085294579723, "significant_q<0.05": false},
          {"residue": "Asp", "residue_full": "Aspartate", "rho": 0.06423181265669871, "p": 0.27561167584494295, "n": 290, "q_bh": 0.32728886506586974, "significant_q<0.05": false},
          {"residue": "Lys", "residue_full": "Lysine", "rho": 0.10504876309297326, "p": 0.07407743793647786, "n": 290, "q_bh": 0.09383142138620529, "significant_q<0.05": false},
          {"residue": "Thr", "residue_full": "Threonine", "rho": 0.209312590655905, "p": 0.000332069344934981, "n": 290, "q_bh": 0.0005257764628137199, "significant_q<0.05": true},
          {"residue": "Pro", "residue_full": "Proline", "rho": 0.23936070727207415, "p": 3.812009245648104e-05, "n": 290, "q_bh": 7.242817566731398e-05, "significant_q<0.05": true},
          {"residue": "Hyp", "residue_full": "Hydroxyproline", "rho": 0.2868495397503328, "p": 6.744542798917197e-07, "n": 290, "q_bh": 1.6018289147428343e-06, "significant_q<0.05": true},
          {"residue": "Glu", "residue_full": "Glutamate", "rho": 0.2878846439253688, "p": 6.122736080529262e-07, "n": 290, "q_bh": 1.6018289147428343e-06, "significant_q<0.05": true},
          {"residue": "Tyr", "residue_full": "Tyrosine", "rho": 0.298630565448188, "p": 2.1921059411262746e-07, "n": 290, "q_bh": 6.941668813566536e-07, "significant_q<0.05": true},
          {"residue": "Leu", "residue_full": "Leucine", "rho": 0.4155214807485684, "p": 1.5595984792017566e-13, "n": 290, "q_bh": 5.926474220966675e-13, "significant_q<0.05": true},
          {"residue": "Phe", "residue_full": "Phenylalanine", "rho": 0.4266379695616228, "p": 2.953033023159118e-14, "n": 290, "q_bh": 1.402690686000581e-13, "significant_q<0.05": true},
          {"residue": "Ile", "residue_full": "Isoleucine", "rho": 0.42925848715892984, "p": 1.9769107861865514e-14, "n": 290, "q_bh": 1.252043497918149e-13, "significant_q<0.05": true},
          {"residue": "Val", "residue_full": "Valine", "rho": 0.48843529342736264, "p": 8.574343291343259e-19, "n": 290, "q_bh": 8.145626126776097e-18, "significant_q<0.05": true}
        ]
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true,
      "notes": "Within ray-finned fish glycine does remain the strongest negative correlate (rho = -0.63). But hydroxyproline and proline fall from 2nd and 3rd strongest positive in the pooled screen (0.44, 0.41) to 7th and 8th (0.29, 0.24), behind Val, Ile, Phe, Leu, Tyr and Glu; they stay FDR-significant here. Within mammals they are 6th and 7th (Pro 0.22, Hyp 0.20) and not FDR-significant (t3-within-mammals-off-reference). In both classes their rho is roughly half to two-thirds of the pooled value. Fish have both lower C:N and lower Hyp and Pro than mammals (medians 3.16 against 3.22, Hyp 72 against 95, Pro 110 against 122 residues/1000), so pooling the classes inflates these two correlations. Part of their prominence in claim 3 is therefore a between-class difference rather than a within-class association."
    },
    {
      "call_id": "t3-dense-family-strict-q-off-reference",
      "tool": "amino_acid_screen",
      "args": {"min_n": 400, "fdr_q": 0.001},
      "select": "/results",
      "expect": {
        "kind": "literal",
        "value": [
          {"residue": "Gly", "residue_full": "Glycine", "rho": -0.6160052561250701, "p": 6.643748814276406e-47, "n": 436, "q_bh": 1.1294372984269892e-45, "significant_q<0.001": true},
          {"residue": "Arg", "residue_full": "Arginine", "rho": -0.3625147599620659, "p": 5.486939172137926e-15, "n": 436, "q_bh": 1.3325423703763536e-14, "significant_q<0.001": true},
          {"residue": "Ser", "residue_full": "Serine", "rho": -0.29674327725788596, "p": 2.5931667815528584e-10, "n": 436, "q_bh": 4.898203920710955e-10, "significant_q<0.001": true},
          {"residue": "Met", "residue_full": "Methionine", "rho": -0.28475650395291513, "p": 1.4699169676732855e-09, "n": 435, "q_bh": 2.4988588450445854e-09, "significant_q<0.001": true},
          {"residue": "Ala", "residue_full": "Alanine", "rho": -0.24080667309555426, "p": 3.6027706363383304e-07, "n": 436, "q_bh": 5.103925068145968e-07, "significant_q<0.001": true},
          {"residue": "His", "residue_full": "Histidine", "rho": -0.22469525206881527, "p": 2.1473433187086214e-06, "n": 436, "q_bh": 2.8080643398497356e-06, "significant_q<0.001": true},
          {"residue": "Thr", "residue_full": "Threonine", "rho": -0.1117970310568797, "p": 0.019542724353701055, "n": 436, "q_bh": 0.020764144625807373, "significant_q<0.001": false},
          {"residue": "Asp", "residue_full": "Aspartate", "rho": 0.09129002719724856, "p": 0.05681929530167535, "n": 436, "q_bh": 0.05681929530167535, "significant_q<0.001": false},
          {"residue": "Lys", "residue_full": "Lysine", "rho": 0.12844090123024954, "p": 0.007245245483200482, "n": 436, "q_bh": 0.00821127821429388, "significant_q<0.001": false},
          {"residue": "Tyr", "residue_full": "Tyrosine", "rho": 0.2074851566342331, "p": 1.285497850928354e-05, "n": 435, "q_bh": 1.560961676127287e-05, "significant_q<0.001": true},
          {"residue": "Phe", "residue_full": "Phenylalanine", "rho": 0.2807609655971062, "p": 2.4306743535465393e-09, "n": 436, "q_bh": 3.7564967282082884e-09, "significant_q<0.001": true},
          {"residue": "Glu", "residue_full": "Glutamate", "rho": 0.3331595370632363, "p": 9.22216423447975e-13, "n": 436, "q_bh": 1.9597098998269467e-12, "significant_q<0.001": true},
          {"residue": "Ile", "residue_full": "Isoleucine", "rho": 0.3786302625167558, "p": 2.6068352781813043e-16, "n": 436, "q_bh": 7.386033288180362e-16, "significant_q<0.001": true},
          {"residue": "Val", "residue_full": "Valine", "rho": 0.4074273167786331, "p": 7.26768579520926e-19, "n": 436, "q_bh": 2.4710131703711483e-18, "significant_q<0.001": true},
          {"residue": "Pro", "residue_full": "Proline", "rho": 0.4112358546898367, "p": 3.1950506917001684e-19, "n": 436, "q_bh": 1.3578965439725716e-18, "significant_q<0.001": true},
          {"residue": "Hyp", "residue_full": "Hydroxyproline", "rho": 0.4433475302003829, "p": 2.0298625294535675e-22, "n": 436, "q_bh": 1.1502554333570216e-21, "significant_q<0.001": true},
          {"residue": "Leu", "residue_full": "Leucine", "rho": 0.49880122289719103, "p": 8.330809054299544e-29, "n": 436, "q_bh": 7.0811876961546125e-28, "significant_q<0.001": true},
          {"residue": "Cys", "residue_full": "Cysteine", "rho": null, "p": null, "n": 110, "q_bh": null, "significant_q<0.001": false},
          {"residue": "Hyl", "residue_full": "Hydroxylysine", "rho": null, "p": null, "n": 347, "q_bh": null, "significant_q<0.001": false}
        ]
      },
      "rel_tolerance": 1e-06,
      "abs_tolerance": 1e-09,
      "enabled": true,
      "notes": "Cys is reported in 110 of 436 rows and Hyl in 347. Excluding both from the family and tightening FDR to q < 0.001 still leaves 14 of 17 residues significant. Only Thr (q = 0.021) and Lys (q = 0.0082) drop out relative to q < 0.05; Asp was never significant (q = 0.064 in the published 19-residue screen, 0.057 in this 17-residue family), so at q < 0.05 this family flags 16 of 17, the same 16 residues the published 19-residue screen flags. The breadth of the screen is not an artefact of sparse residues or of a lenient threshold at n = 436."
    }
  ]
}