{
 "dataset": {
  "archive": "dandi",
  "id": "000773",
  "label": "DANDI:000773"
 },
 "page": "datasets/dandi/000773/",
 "generated": "2026-11-05T07:41:23+00:00",
 "caution": "A failure to reproduce is not a finding that the paper is wrong. It may be a failure of the attempt.",
 "outcome_counts": {
  "reproduced": 2,
  "partially_reproduced": 1,
  "not_reproduced": 0,
  "blocked": 3
 },
 "outcome_counts_with_outside_data": {
  "reproduced": 1,
  "partially_reproduced": 2,
  "not_reproduced": 5,
  "blocked": 1
 },
 "outcome_counts_elsewhere": {
  "reproduced": 5,
  "partially_reproduced": 0,
  "not_reproduced": 0,
  "blocked": 1
 },
 "outcome_counts_links": {
  "reproduced": 0,
  "partially_reproduced": 1,
  "not_reproduced": 0,
  "blocked": 0
 },
 "attempts": [
  {
   "id": "000773-hart-2024-01",
   "dataset": {
    "archive": "dandi",
    "id": "000773",
    "version": "0.261004.1511",
    "doi": "10.48324/dandi.000773/0.261004.1511"
   },
   "key": "dandi:000773",
   "label": "DANDI:000773",
   "dataset_version": "0.261004.1511",
   "paper": "10.1016/j.cub.2024.05.031",
   "paper_title": "Pheromone representation in the ant antennal lobe changes with age",
   "date": "2026-10-31",
   "kind": "agent",
   "accountable": "CatalystNeuro",
   "agent": {
    "harness": "Claude Code",
    "model": "claude-opus-5-5",
    "protocol": "pilot-12"
   },
   "runner": {
    "platform": "GitHub Actions",
    "url": "https://github.com/catalystneuro/encore/actions/runs/41222502726"
   },
   "tier": "agent_verified",
   "deprecated": "",
   "deprecated_on": null,
   "entry": "000773-hart-2024",
   "entry_title": "Age and odor coding in the clonal raider ant antennal lobe (Hart et al. 2024)",
   "description": "Tests the findings of Hart, Lopes, Frank and Kronauer (2024) on dandiset 000773, which holds the two-photon calcium imaging of the right antennal lobe of 21 young and old GCaMP6s ants stimulated with two alarm pheromones and five general odorants. It computes the whole-lobe peak response maps, counts responding ROIs, and measures the six focal glomeruli, located with the paper's supplementary traces (Data S2). The behaviour claims are tested on the paper's Data S1.",
   "notebooks": [
    {
     "path": "reproduce_hart_2024.ipynb",
     "title": "Reproducing Hart et al. (2024) from dandiset 000773",
     "executed": "notebooks/000773-hart-2024/reproduce_hart_2024.html"
    }
   ],
   "outcomes": [
    {
     "claim": "fig1c-old-faster-alarm",
     "statement": "For the pooled alarm pheromone stimuli, separate logistic curves for young and old ants fit better than a single curve, and old ants leave the nest faster.",
     "source": "Results, paragraph 1; Figure 1C; Table S1, row \"Pooled alarm pheromone stimuli\"",
     "quote": "For the pooled alarm pheromone stimuli, the preferred model incorporated separate curves for young and old ants, with old ants responding faster",
     "statistics": [
      {
       "test": "extra sum-of-squares F test, single against separate logistic curves",
       "statistic": "F(3, 314) = 8.006",
       "p": 0.0001,
       "p_relation": "<",
       "n": "20 trials (6 with 4-methyl-3-heptanone, 7 with 4-methyl-3-heptanol, 7 with the blend)"
      }
     ],
     "kind": "finding",
     "inferred": "",
     "outcome": "reproduced",
     "blocked_by": null,
     "notebook": "reproduce_hart_2024.ipynb",
     "notes": "Separate logistic curves for young and old ants are preferred, F(3, 314) = 8.006, p = 3.7e-05, the values of Table S1. All six parameters are within tolerance. Old ants are faster under both readings of the criterion; K is 0.0986 for old and 0.0695 for young ants, and the old curve is above the young one at every second from 0 to 60 s (by at least 0.061).",
     "deviations": [
      "Tested on Data S1, which the dandiset does not hold (data elsewhere). Starting values of the fits (YM 0.9, Y0 0.5, K 0.05) are a standard choice; Prism's are not given."
     ],
     "assumed": false,
     "disagreed": false,
     "data": "elsewhere",
     "where": "Data S1, supplemental information of doi:10.1016/j.cub.2024.05.031 (PMC11265976)",
     "proposed": null
    },
    {
     "claim": "fig1c-vehicle-single-curve",
     "statement": "For the vehicle control a single curve for young and old ants is preferred, so the two ages have similar baseline behavior.",
     "source": "Results, paragraph 1; Figure 1C right; Table S1, row \"Vehicle control\"",
     "quote": "For the vehicle control, a single curve for young and old ants was the preferred model (extra sum-of-squares F test), indicating similar baseline behavioral patterns",
     "statistics": [
      {
       "test": "extra sum-of-squares F test, single against separate quadratic curves",
       "statistic": "F(3, 90) = 2.617",
       "p": 0.056,
       "n": "6 vehicle trials (from the 90 residual degrees of freedom at 8 time points per age)"
      }
     ],
     "kind": "finding",
     "inferred": "",
     "outcome": "reproduced",
     "blocked_by": null,
     "notebook": "reproduce_hart_2024.ipynb",
     "notes": "The quadratic fits give F(3, 90) = 2.617, p = 0.056, as in Table S1, so the single curve is not rejected.",
     "deviations": [
      "Tested on Data S1 (data elsewhere)."
     ],
     "assumed": false,
     "disagreed": false,
     "data": "elsewhere",
     "where": "Data S1, supplemental information of doi:10.1016/j.cub.2024.05.031 (PMC11265976)",
     "proposed": null
    },
    {
     "claim": "fig1b-alarm-stimuli-similar",
     "statement": "The three alarm stimuli (4-methyl-3-heptanone, 4-methyl-3-heptanol and the 9:1 blend) make the ants leave the nest in a similar manner; one curve fits all three.",
     "source": "Results, paragraph 1; Figure 1B; Table S1, row \"Alarm pheromone stimuli, comparing stimuli\"",
     "quote": "all three alarm pheromone stimuli caused ants to leave the nest in a similar manner, so these samples were pooled for analyzing effects of age",
     "statistics": [
      {
       "test": "extra sum-of-squares F test, single against separate logistic curves",
       "statistic": "F(6, 151) = 2.144",
       "p": 0.052,
       "n": "20 trials (6, 7 and 7)"
      }
     ],
     "kind": "finding",
     "inferred": "",
     "outcome": "reproduced",
     "blocked_by": null,
     "notebook": "reproduce_hart_2024.ipynb",
     "notes": "One logistic curve for the three alarm stimuli is not rejected, F(6, 151) = 2.144, p = 0.052, as in Table S1.",
     "deviations": [
      "Tested on Data S1 (data elsewhere)."
     ],
     "assumed": false,
     "disagreed": false,
     "data": "elsewhere",
     "where": "Data S1, supplemental information of doi:10.1016/j.cub.2024.05.031 (PMC11265976)",
     "proposed": null
    },
    {
     "claim": "figs1-heptanone-separate-curves",
     "statement": "For 4-methyl-3-heptanone alone, separate curves for young and old ants are preferred, with old ants above young.",
     "source": "Figure S1 and legend; Table S1, row \"4-methyl-3-heptanone, comparing age\"",
     "quote": "For 4-methyl-3-heptanone, a model with separate curves was preferred.",
     "statistics": [
      {
       "test": "extra sum-of-squares F test, single against separate logistic curves",
       "statistic": "F(3, 90) = 6.377",
       "p": 0.0006,
       "n": "6 trials"
      }
     ],
     "kind": "finding",
     "inferred": "",
     "outcome": "reproduced",
     "blocked_by": null,
     "notebook": "reproduce_hart_2024.ipynb",
     "notes": "Separate curves are preferred for 4-methyl-3-heptanone, F(3, 90) = 6.377, p = 0.00057 (Table S1 F(3, 90) = 6.377). The old curve is above the young one at every second from 0 to 60 s (reading b); K is also slightly higher for old ants (0.0973 against 0.0949). Fitted YM and Y0 are 0.830 and 0.534 for young and 0.921 and 0.662 for old ants, as Table S1.",
     "deviations": [
      "Tested on Data S1 (data elsewhere)."
     ],
     "assumed": false,
     "disagreed": false,
     "data": "elsewhere",
     "where": "Data S1, supplemental information of doi:10.1016/j.cub.2024.05.031 (PMC11265976)",
     "proposed": null
    },
    {
     "claim": "figs1-heptanol-blend-single-curve",
     "statement": "For 4-methyl-3-heptanol and for the blend, single curves for both ages are preferred, though the trends are similar to 4-methyl-3-heptanone.",
     "source": "Figure S1 and legend; Table S1, rows \"4-methyl-3-heptanol\" and \"Blend, comparing age\"",
     "quote": "For 4-methyl-3-heptanol and the blend, similar trends were observed, but single curve models were preferred",
     "statistics": [
      {
       "about": "4-methyl-3-heptanol, young against old",
       "test": "extra sum-of-squares F test",
       "statistic": "F(3, 106) = 1.979",
       "p": 0.122,
       "n": "7 trials"
      },
      {
       "about": "blend, young against old",
       "test": "extra sum-of-squares F test",
       "statistic": "F(3, 106) = 2.529",
       "p": 0.061,
       "n": "7 trials"
      }
     ],
     "kind": "finding",
     "inferred": "",
     "outcome": "reproduced",
     "blocked_by": null,
     "notebook": "reproduce_hart_2024.ipynb",
     "notes": "Single curves are not rejected for 4-methyl-3-heptanol (F(3, 106) = 1.979, p = 0.12) or the blend (F(3, 106) = 2.529, p = 0.061), the values of Table S1. In both the fitted old curve is above the young one from 0 to 60 s, the similar trend the legend describes.",
     "deviations": [
      "Tested on Data S1 (data elsewhere)."
     ],
     "assumed": false,
     "disagreed": false,
     "data": "elsewhere",
     "where": "Data S1, supplemental information of doi:10.1016/j.cub.2024.05.031 (PMC11265976)",
     "proposed": null
    },
    {
     "claim": "fig2b-gcamp-baseline-unchanged",
     "statement": "Total GCaMP6s fluorescence in the AL of dissected brains does not differ between young and old ants.",
     "source": "Results, paragraph 3; Figure 2A and 2B",
     "quote": "Total fluorescence intensity did not differ with age, and calcium responses measured by changes in GCaMP6s fluorescence should therefore be comparable between the two age cohorts",
     "statistics": [
      {
       "test": "Welch's t test",
       "p": 0.21
      }
     ],
     "kind": "finding",
     "inferred": "",
     "outcome": "blocked",
     "blocked_by": "other",
     "notebook": null,
     "notes": "The confocal stacks are not in the dandiset; they are in the Brain Image Library (doi:10.35077/g.1153). The notebook reads the landing page, which lists 15 brains (7 young, 8 old) under download.brainimagelibrary.org and carries two notices, printed (the repository \"is under review for potential modification in compliance with Administration directives\", and \"BIL data will be unavailable between October 5 and October 16\"). The notebook then tries the download server three times, 30 s apart with a 60 s timeout each, and every try timed out (ReadTimeout, ReadTimeout, ConnectTimeout in the final run's output; three ReadTimeouts in the round 2 run). Requests made from the shell during the attempt also timed out but are in no output. The data were never read, so the block is other (the server was not reached), not data_unreadable. The number of brains could not be compared with Figure 2B.",
     "deviations": [],
     "assumed": false,
     "disagreed": false,
     "data": "elsewhere",
     "where": "Brain Image Library, doi:10.35077/g.1153",
     "proposed": null
    },
    {
     "claim": "fig2d-whole-al-alarm-no-age-effect",
     "statement": "The mean calcium response across the entire AL to each alarm pheromone does not differ between young and old ants, at 3% or at 48%.",
     "source": "Results, paragraph 4; Figure 2C and 2D",
     "quote": "we observed no effect of age on mean calcium response across the entire AL for either of the two tested concentrations",
     "statistics": [
      {
       "test": "Welch's t tests, one per pheromone and concentration",
       "p": 0.05,
       "p_relation": ">",
       "n": "3 young and 4 old ants at 3%, 7 young and 7 old at 48%"
      }
     ],
     "kind": "finding",
     "inferred": "",
     "outcome": "reproduced",
     "blocked_by": null,
     "notebook": "reproduce_hart_2024.ipynb",
     "notes": "None of the four Welch tests finds an age difference in the whole-AL mean peak fold change. At 3%, p = 0.78 (4-methyl-3-heptanone) and 0.065 (4-methyl-3-heptanol), 3 young and 4 old ants; at 48%, p = 0.39 and 0.14, 7 and 7 ants. The outline shrunk or grown by 5 pixels and the 32-plane projection give the same (smallest p 0.061). Cohort means are -0.15 to -0.26; the minimum filter makes a non-responding AL negative (paraffin oil median -0.23).",
     "deviations": [
      "The AL outline is found by Otsu's threshold on the mean raw maximum projection, as the criterion allows; the paper drew it by hand.",
      "Frames are aligned by whole-pixel phase correlation to the mean of frames 1-5, a stand-in for the ImageJ Image Stabilizer. The median shift is 1 pixel; 2742 of 167,706 frame estimates above 20 pixels were taken as failures and not applied."
     ],
     "assumed": false,
     "disagreed": false,
     "data": "dataset",
     "where": null,
     "proposed": null
    },
    {
     "claim": "figs3b-whole-al-general-decline",
     "statement": "At 48%, 3-hexanone and isopropanol evoke broad AL responses in young but not old ants; mean peak responses across the AL fall more than 9-fold (3-hexanone) and more than 57-fold (isopropanol) with age.",
     "source": "Results, paragraph 7; Figure S3A and S3B",
     "quote": "Mean peak calcium responses across the AL declined dramatically with age for 3-hexanone (>9-fold decline) and isopropanol (>57-fold decline)",
     "statistics": [
      {
       "about": "3-hexanone, 48%, young against old",
       "test": "Welch's t test",
       "p": 5.1e-06,
       "n": "7 young and 7 old ants"
      },
      {
       "about": "isopropanol, 48%, young against old",
       "test": "Welch's t test",
       "p": 0.0068,
       "n": "7 young and 7 old ants"
      }
     ],
     "kind": "finding",
     "inferred": "",
     "outcome": "partially_reproduced",
     "blocked_by": null,
     "notebook": "reproduce_hart_2024.ipynb",
     "notes": "At 48% the whole-AL response falls with age for 3-hexanone (young 1.613, old 0.039, t = 7.39, p = 2.0e-05) and isopropanol (young 0.860, old -0.151, p = 0.0085). The 3-hexanone ratio of cohort means, 41.0, exceeds the 8.1 needed (24.0 and 300 with the outline shrunk or grown). The isopropanol old mean is negative, so its ratio (-5.68) cannot express a fold decline and does not reach 51.3. The measure has a negative offset, from the minimum filter on noisy per-pixel fold change: paraffin oil trials give a median of -0.23 (per ant -0.45 to 0.075). With each ant's paraffin oil value subtracted, both declines hold (p = 5.3e-05 and 0.012) and both old means are positive, but the ratios are 6.0 and 9.5, below 8.1 and 51.3; so the ratios are not reached either way, and the outcome does not rest on the offset. Without the minimum filter the map saturates near the clip value of 4 even for paraffin oil (median 3.80), so that variant measures no response. The paper's Figure S3B values are not in the text, so the offset could not be compared with its scale.",
     "deviations": [
      "As fig2d-whole-al-alarm-no-age-effect (outline by threshold, stabilisation by phase correlation).",
      "Variants added in revision, to show the offset: each ant's paraffin oil value subtracted, and the map without the minimum filter."
     ],
     "assumed": false,
     "disagreed": false,
     "data": "dataset",
     "where": null,
     "proposed": null
    },
    {
     "claim": "figs3b-whole-al-other-no-age-effect",
     "statement": "The age decline of whole-AL responses is specific to 3-hexanone and isopropanol at 48%; ethylpyrazine, ethanol and propionic acid at 48%, and all five general odorants at 3%, show no age effect.",
     "source": "Results, paragraph 7; Figure S3B",
     "quote": "an effect that was specific to these two odorants when presented at 48% concentration",
     "statistics": [
      {
       "test": "Welch's t tests, one per odorant and concentration",
       "p": 0.05,
       "p_relation": ">",
       "n": "3 young and 4 old ants at 3%, 7 and 7 at 48%"
      }
     ],
     "kind": "finding",
     "inferred": "",
     "outcome": "reproduced",
     "blocked_by": null,
     "notebook": "reproduce_hart_2024.ipynb",
     "notes": "The eight tests (five general odorants at 3%; ethylpyrazine, ethanol and propionic acid at 48%) give p from 0.070 to 0.74; the smallest over the outline, projection and paraffin-subtracted variants is 0.062 (ethylpyrazine at 48%). The map without the minimum filter, saturated near 4 for every odor and paraffin oil, gives propionic acid p = 0.042 at 48%; it measures no response and is not a test of the claim.",
     "deviations": [
      "As fig2d-whole-al-alarm-no-age-effect."
     ],
     "assumed": false,
     "disagreed": false,
     "data": "dataset",
     "where": null,
     "proposed": null
    },
    {
     "claim": "fig2e-heptanone-breadth-falls",
     "statement": "At 48%, 4-methyl-3-heptanone activates a few more glomeruli in young than in old ants (medians of 4 against 2 responding ROIs), while the count for 4-methyl-3-heptanol does not differ with age.",
     "source": "Results, paragraph 4; Figure 2E",
     "quote": "While there was no significant difference for 4-methyl-3-heptanol, 4-methyl-3-heptanone activated a few more glomeruli in young ants than in old ants (medians of 4 versus 2 responding ROIs)",
     "statistics": [
      {
       "about": "4-methyl-3-heptanone, young against old",
       "test": "Wilcoxon test",
       "p": 0.05,
       "p_relation": "<",
       "n": "7 young and 7 old ants"
      },
      {
       "about": "4-methyl-3-heptanol, young against old",
       "test": "Wilcoxon test",
       "p": 0.05,
       "p_relation": ">",
       "n": "7 young and 7 old ants"
      }
     ],
     "kind": "finding",
     "inferred": "",
     "outcome": "blocked",
     "blocked_by": "methods_underspecified",
     "notebook": null,
     "notes": "The paper drew glomerulus-sized ROIs by hand and does not give their size, positions or the per-ant counts; Data S2 holds traces of the six focal glomeruli only, no counts, so the criterion's comparison with \"the paper's counts in Data S2\" cannot be made. The claims file's automated ROIs were run at three glomerulus diameters, and the outcome depends on the size. For 4-methyl-3-heptanone the medians are 4 (young) and 2 (old) at 25 pixels (p = 0.22), 1 and 0 at 35 pixels (p = 0.35), 0 and 0 at 50 pixels (p = 0.71), so partially reproduced at the first two and not reproduced at the third; none finds the significant difference the paper reports. For 4-methyl-3-heptanol no size finds a difference (p 0.23 to 0.95). Reading (b) of the motion exclusions now leaves out a trial that Data S2 lacks for one glomerulus while it holds that odor's other repeats (3 trials, one in each of three ants); it gives p = 0.17, 0.28 and 0.60 for 4-methyl-3-heptanone. Data S2 lacks no trial for every glomerulus, so the round 1 reading (b) equalled reading (a). The question is for the paper's authors (see findings).",
     "deviations": [
      "ROIs by the claims file's automated rule (local maxima at >= 0.2 recurring in two or more trials, disks of one glomerulus diameter), at diameters of 25, 35 and 50 pixels; the paper drew them by hand."
     ],
     "assumed": false,
     "disagreed": false,
     "data": "dataset",
     "where": null,
     "proposed": null
    },
    {
     "claim": "fig2-alarm-sparse",
     "statement": "Alarm pheromones are represented sparsely at both ages; only a small fraction of the roughly 500 AL glomeruli respond to them.",
     "source": "Results, paragraphs 4 and 5; Figure 2C and 2E",
     "quote": "Calcium responses to 4-methyl-3-heptanone and 4-methyl-3-heptanol were sparse at both tested ages ... Out of the ~500 glomeruli in the ant AL, only a small fraction responded to alarm pheromones in young and/or old ants",
     "statistics": [],
     "kind": "finding",
     "inferred": "",
     "outcome": "blocked",
     "blocked_by": "methods_underspecified",
     "notebook": null,
     "notes": "Depends on fig2e-heptanone-breadth-falls, which is blocked. The counts are small at every size; the largest count of any ant for 4-methyl-3-heptanone is 32, 17 and 7 ROIs under reading (a) and 39, 23 and 9 under reading (b) (25, 35 and 50 pixels), and for 4-methyl-3-heptanol 11, 4 and 2. The 10% reading (50 ROIs) holds at every size; the 5% reading (25 ROIs) fails at 25 pixels.",
     "deviations": [],
     "assumed": false,
     "disagreed": false,
     "data": "dataset",
     "where": null,
     "proposed": null
    },
    {
     "claim": "figs3c-general-breadth-no-age-effect",
     "statement": "Age does not change the number of glomeruli activated by ethylpyrazine, ethanol or propionic acid at 48%.",
     "source": "Results, paragraph 7; Figure S3C",
     "quote": "we found no effect of age on the number of activated glomeruli for ethylpyrazine, ethanol, or propionic acid",
     "statistics": [
      {
       "test": "Wilcoxon tests, one per odorant",
       "p": 0.05,
       "p_relation": ">",
       "n": "7 young and 7 old ants"
      }
     ],
     "kind": "finding",
     "inferred": "",
     "outcome": "blocked",
     "blocked_by": "methods_underspecified",
     "notebook": null,
     "notes": "Blocked for the reason fig2e-heptanone-breadth-falls is. By the letter of the criterion no odorant shows an age difference in the count of responding ROIs (smallest p 0.075, ethanol at 35 pixels; ethylpyrazine 0.083 to 0.32, propionic acid 0.36 to 0.94), but the counts come from the claims file's automated ROIs, which do not find the age effect the paper reports for 4-methyl-3-heptanone and whose counts are implausible as glomerulus counts (17 to 173 ROIs per ant at 25 pixels; up to 128 ROIs responding to ethanol at 25 pixels and 72 in one young ant at 35 pixels). A null result from a procedure that misses the paper's effect is not a test. Changed from reproduced (assumed) after the round 1 verifier (problem 7).",
     "deviations": [
      "ROIs as in fig2e-heptanone-breadth-falls, at three diameters; the paper drew them by hand and gives no positions or counts."
     ],
     "assumed": false,
     "disagreed": true,
     "data": "dataset",
     "where": null,
     "proposed": null
    },
    {
     "claim": "data-s2-glomerulus-link",
     "statement": "The per-glomerulus fold-change traces of Data S2 come from the dandiset's ants and trials, and each focal glomerulus can be found in the dandiset's imaging where the per-trial responses follow Data S2's, so that the focal glomerulus claims can be measured on the dandiset.",
     "source": "Data S2 (NIHMS1997610-supplement-MMC3.xlsx, PMC11265976.1); STAR Methods, Image processing, paragraph on focal glomeruli",
     "quote": "",
     "statistics": [],
     "kind": "link",
     "inferred": "",
     "outcome": "partially_reproduced",
     "blocked_by": null,
     "notebook": "reproduce_hart_2024.ipynb",
     "notes": "All 14 Data S2 ants at 48% are dandiset ants at 48%, and Data S2's age labels agree with the dandiset's ages (P16 to P20 and P60 to P61). At the place found for each glomerulus from all its Data S2 trials (the criterion), 71 of 108 ant, glomerulus and set triples (66%) have per-trial AUCs that correlate with Data S2's at r >= 0.7, short of the 80% for the link to hold. Young ants agree in 46 of 56, old ants in 25 of 52; the main set in 57 of 81 and the Figure 3C and 3D set in 14 of 27. Over all 1672 trials r = 0.793. Two checks added in revision show this is optimistic. Against the best r reached with Data S2's values shuffled (200 shuffles per placement), all 42 young placements are above chance (p < 0.05) but only 24 of 39 old ones (18 of 39 with the odors shuffled as blocks); triples agree at 66 of 90 placements above chance and 5 of 18 below. Read where the glomerulus was placed without that odor's trials, 44 of 108 triples agree (41%; young 35 of 56, old 9 of 52; Figure 3C and 3D set 4 of 27), r = 0.731 over all trials; alarm trials read where the glomerulus was placed from the general odorants alone agree in 19 of 79 (old 4 of 37). Held out, the link would fail; in old ants it fails either way. The dependent claims are tested; their old-ant values, PGb with the alarm pheromones most of all, are where the link is weakest.",
     "deviations": [
      "The positions are chosen by correlation with Data S2's per-trial peak responses, so under the criterion the dandiset's values are not independent of Data S2's pattern across trials; amplitudes are not matched. A held-out placement and a permutation chance level are shown."
     ],
     "assumed": false,
     "disagreed": false,
     "data": "with_inputs",
     "where": null,
     "proposed": null
    },
    {
     "claim": "figs2a-mean-six-glomeruli-no-age-effect",
     "statement": "The alarm pheromone response averaged over the six focal glomeruli does not differ between young and old ants.",
     "source": "Results, paragraph 5; Figure S2A",
     "quote": "The mean of responses across these six glomeruli did not differ with age",
     "statistics": [
      {
       "test": "Welch's t tests, one per pheromone",
       "p": 0.05,
       "p_relation": ">",
       "n": "7 young and 7 old ants, 48%"
      }
     ],
     "kind": "finding",
     "inferred": "",
     "outcome": "reproduced",
     "blocked_by": null,
     "notebook": "reproduce_hart_2024.ipynb",
     "notes": "The mean AUC over the glomeruli identified in each ant does not differ with age. At the held-out placement (reading a, radius 8) p = 0.58 (4-methyl-3-heptanone) and 0.23 (4-methyl-3-heptanol); over every variant, including the placement from all trials (p = 0.81 and 0.25) and placements above chance only, p >= 0.0799 (general odorants only, reading b, 4-methyl-3-heptanone). Data S2's own values give p = 0.14 and 0.053. The partial link (data-s2-glomerulus-link) bears on the old-ant values. This null result is weak evidence, since it comes from the same located ROIs that miss the paper's positive old-ant effects (fig3b, fig3d), and a null from a procedure that misses the paper's effects is a weak test, the reason figs3c-general-breadth-no-age-effect is blocked. It is kept as reproduced because the glomeruli are placed from Data S2 per ant (above chance in every young ant), not by the automated rule of figs3c, and no variant reaches p < 0.05.",
     "deviations": [
      "Glomerulus positions are not in the dandiset; each is placed where the dandiset's per-trial responses best correlate with Data S2's (data-s2-glomerulus-link), so the test uses Data S2 (data with_inputs).",
      "Assumed the held-out placement as the main measure: each odor's AUCs are read where the glomerulus was placed from Data S2's trials of the other odors (leave one odor out), so that the tested values did not choose the place (verifier, round 1, problem 2). The placement from all trials (round 1's main measure), placements above chance only, and for the alarm pheromones the placement from the general odorants alone are shown as variants.",
      "Assumed a disk ROI of 8 pixels radius (2.3 um) on the three-plane projection; the paper gives no ROI size. Radii 5 and 12 are shown.",
      "Assumed fold-change reading (a), the ROI mean of the projected signal turned into fold change, the usual ImageJ order for \"a max Z-projection was generated ... and the fluorescence fold change was calculated\" (STAR Methods, Image processing); reading (b) is shown, leaving out pixels whose baseline F0 is at or below the ant's background level (the median of the raw maximum projection outside the AL), an assumption, since a pixel at background level divides by noise.",
      "Per ant, the mean over the trials Data S2 holds for that glomerulus (the paper's own trial selection); all trials are shown as a variant.",
      "The 40-frame traces are aligned with one set of shifts per trial, measured on the middle plane, not per plane."
     ],
     "assumed": true,
     "disagreed": false,
     "data": "with_inputs",
     "where": null,
     "proposed": null
    },
    {
     "claim": "fig3b-g2-heptanone-weaker-old",
     "statement": "In glomeruli G1 to G4 alarm pheromone responses tend to weaken with age; the decline is significant only for 4-methyl-3-heptanone in G2.",
     "source": "Results, paragraph 5; Figure 3B; Figure S2B",
     "quote": "For G1\u2013G4, we observed a trend toward weaker responses with increasing age, which was statistically significant for responses to 4-methyl-3-heptanone in G2",
     "statistics": [
      {
       "about": "G2, 4-methyl-3-heptanone, young against old",
       "test": "Welch's t test",
       "p": 0.0068,
       "n": "7 young and 7 old ants"
      },
      {
       "about": "the other seven glomerulus and pheromone pairs of G1 to G4",
       "test": "Welch's t tests",
       "p": 0.05,
       "p_relation": ">"
      }
     ],
     "kind": "finding",
     "inferred": "",
     "outcome": "not_reproduced",
     "blocked_by": null,
     "notebook": "reproduce_hart_2024.ipynb",
     "notes": "At the held-out placement (reading a, radius 8) G2 with 4-methyl-3-heptanone does not differ (young -1.98, old -2.34, p = 0.86), and G1 with 4-methyl-3-heptanone does (8.71 against -1.97, p = 0.033). Over the variants G2 p ranges from 0.0496 to 0.86; only the placement from all trials restricted to placements above chance (5 old ants) gives G2 p < 0.05 (p = 0.0496), with one other pair also significant, which would be partially reproduced. At the placement from all trials (round 1) p = 0.24. Young means exceed old ones in 5 of 8 pairs (held out). Data S2's own values give G2 p = 0.0069 and no other pair below 0.05, as the paper. The outcome rests on the located ROIs; the link is partial, weakest in old ants.",
     "deviations": [
      "Glomerulus positions are not in the dandiset; each is placed where the dandiset's per-trial responses best correlate with Data S2's (data-s2-glomerulus-link), so the test uses Data S2 (data with_inputs).",
      "Assumed the held-out placement as the main measure: each odor's AUCs are read where the glomerulus was placed from Data S2's trials of the other odors (leave one odor out), so that the tested values did not choose the place (verifier, round 1, problem 2). The placement from all trials (round 1's main measure), placements above chance only, and for the alarm pheromones the placement from the general odorants alone are shown as variants.",
      "Assumed a disk ROI of 8 pixels radius (2.3 um) on the three-plane projection; the paper gives no ROI size. Radii 5 and 12 are shown.",
      "Assumed fold-change reading (a), the ROI mean of the projected signal turned into fold change, the usual ImageJ order for \"a max Z-projection was generated ... and the fluorescence fold change was calculated\" (STAR Methods, Image processing); reading (b) is shown, leaving out pixels whose baseline F0 is at or below the ant's background level (the median of the raw maximum projection outside the AL), an assumption, since a pixel at background level divides by noise.",
      "Per ant, the mean over the trials Data S2 holds for that glomerulus (the paper's own trial selection); all trials are shown as a variant.",
      "The 40-frame traces are aligned with one set of shifts per trial, measured on the middle plane, not per plane."
     ],
     "assumed": true,
     "disagreed": false,
     "data": "with_inputs",
     "where": null,
     "proposed": null
    },
    {
     "claim": "fig3d-pg-stronger-old",
     "statement": "PGb and PGa respond significantly more strongly to alarm pheromones in old than in young ants; PGa does not respond to 4-methyl-3-heptanone.",
     "source": "Results, paragraph 5; Figure 3C and 3D",
     "quote": "The patterns in PG b and PG a were quite different, with significantly stronger calcium responses in old ants compared to young ants",
     "statistics": [
      {
       "test": "Welch's t tests",
       "p": 0.05,
       "p_relation": "<",
       "n": "PGb 7 young and 7 old ants, PGa 7 young and 6 old"
      }
     ],
     "kind": "finding",
     "inferred": "",
     "outcome": "not_reproduced",
     "blocked_by": null,
     "notebook": "reproduce_hart_2024.ipynb",
     "notes": "At the held-out placement (reading a, radius 8) no pair differs: PGb with 4-methyl-3-heptanone -0.21 (young) against 8.46 (old), p = 0.14; PGb with 4-methyl-3-heptanol p = 0.43; PGa with 4-methyl-3-heptanol -4.21 against 0.08, p = 0.24. No variant meets either reading. The pair significant at the placement from all trials in round 1, PGa with 4-methyl-3-heptanol (11.6 against -1.35, p = 0.044), is not significant at any held-out placement (p 0.13 to 0.33), so it came from choosing the place on the tested trials. With the placement from the general odorants alone, PGb with 4-methyl-3-heptanone is higher in old ants (p = 0.026) but PGa is not (p = 0.33). PGa's 4-methyl-3-heptanone AUC is -3.32 (young) and 1.38 (old), no response. Data S2's own values meet both readings (PGb p = 0.036 and 0.0036, PGa 0.016). At the held-out placement old ants' PGb AUCs are below Data S2's in 6 of 7 ants for 4-methyl-3-heptanone and 7 of 7 for 4-methyl-3-heptanol, the Figure 3C and 3D set where the link is weakest.",
     "deviations": [
      "Glomerulus positions are not in the dandiset; each is placed where the dandiset's per-trial responses best correlate with Data S2's (data-s2-glomerulus-link), so the test uses Data S2 (data with_inputs).",
      "Assumed the held-out placement as the main measure: each odor's AUCs are read where the glomerulus was placed from Data S2's trials of the other odors (leave one odor out), so that the tested values did not choose the place (verifier, round 1, problem 2). The placement from all trials (round 1's main measure), placements above chance only, and for the alarm pheromones the placement from the general odorants alone are shown as variants.",
      "Assumed a disk ROI of 8 pixels radius (2.3 um) on the three-plane projection; the paper gives no ROI size. Radii 5 and 12 are shown.",
      "Assumed fold-change reading (a), the ROI mean of the projected signal turned into fold change, the usual ImageJ order for \"a max Z-projection was generated ... and the fluorescence fold change was calculated\" (STAR Methods, Image processing); reading (b) is shown, leaving out pixels whose baseline F0 is at or below the ant's background level (the median of the raw maximum projection outside the AL), an assumption, since a pixel at background level divides by noise.",
      "Per ant, the mean over the trials Data S2 holds for that glomerulus (the paper's own trial selection); all trials are shown as a variant.",
      "The 40-frame traces are aligned with one set of shifts per trial, measured on the middle plane, not per plane."
     ],
     "assumed": true,
     "disagreed": false,
     "data": "with_inputs",
     "where": null,
     "proposed": null
    },
    {
     "claim": "fig3d-pg-60-vs-90-days-same",
     "statement": "PGb and PGa responses to alarm pheromones do not differ between 60-day-old ants and the 90-day-old ants of the earlier study, so the change occurs between 14 and 60 days.",
     "source": "Results, paragraph 6; Figure 3C and 3D",
     "quote": "We found no significant differences between 60- and 90-day-old ants",
     "statistics": [
      {
       "test": "Welch's t tests",
       "p": 0.05,
       "p_relation": ">"
      }
     ],
     "kind": "finding",
     "inferred": "",
     "outcome": "blocked",
     "blocked_by": "methods_underspecified",
     "notebook": null,
     "notes": "The readings give different outcomes and neither is a fair test here. Reading (a), both cohorts measured here at the placement from all trials (60-day ants of this dandiset, 90-day ants of DANDI 000467), finds no difference (PGb with 4-methyl-3-heptanone 5.15 against 11.59, p = 0.15; PGb with 4-methyl-3-heptanol p = 0.94; PGa p = 0.41; every variant p >= 0.059). Added after the round 2 verifier, the 60-day values at the held-out placement of the main measure against the same 90-day values also show no difference (8.46 against 11.59, p = 0.60; p = 0.53; PGa 0.08 against 6.53, p = 0.18; reading b p = 0.33 to 0.74); at the placement from the general odorants alone only reading (b) of PGb with 4-methyl-3-heptanone differs (1.74 against 14.14, p = 0.041; reading a p = 0.063). But the 90-day PGa positions are placed from six trials and only 1 of 8 is above chance (r 0.996 to 0.9997, 95th percentiles of the shuffled maxima about 0.999); the PGb positions (12 trials) are above chance in 7 of 8. No held-out placement can be made for the 90-day PGa, whose Data S2 trials are only the six alarm trials. Reading (b), the 60-day AUCs measured here against Data S2's 90-day values, finds differences (held-out placement p = 0.014, 0.022 and 0.016; placement from all trials p = 0.0002, 0.053 and 0.41), but the AUCs measured here are several times smaller than Data S2's for the same ants (60-day PGb with 4-methyl-3-heptanone 5.15 against 20.66; 90-day 11.59 in 000467 against 24.99), so it compares two pipelines, not two ages. With Data S2's values for both ages none differs (p = 0.44, 0.80 and 0.78), but that uses no dandiset data. In 000467 the 8 Data S2 ants at 48% are all present, 12 of 16 ant and glomerulus pairs agree with Data S2 at r >= 0.7, and 95 of 96 trial numbers named in Data S2 match the dandiset's trial of that odor and repeat. Changed from reproduced (assumed reading a) after the round 1 verifier (problems 3 and 4); whether reading (b) should take both ages from Data S2 is asked of a reader.",
     "deviations": [
      "The 90-day ants are located in DANDI 000467 as the 60-day ants are in this dandiset (PGb from 12 Data S2 trials with four odors, PGa from 6 alarm trials), radius 8; the 15 120-frame trials of 000467 (6-methyl-5-hepten-2-one) are left out.",
      "Both readings of the claims file were run; the claim is blocked because they disagree and each has a flaw that no choice in the paper settles."
     ],
     "assumed": false,
     "disagreed": true,
     "data": "with_inputs",
     "where": null,
     "proposed": null
    },
    {
     "claim": "figs4a-general-odorants-in-focal-glomeruli",
     "statement": "3-hexanone excites all six focal glomeruli in young ants and PGb, PGa and G4 in old ants; isopropanol excites all but G1 in young ants only; propionic acid excites G4 at both ages; no other general odorant excites a focal glomerulus.",
     "source": "Results, paragraph 8; Figure S4A",
     "quote": "We found that 3-hexanone activated all six glomeruli in young ants as well as activating PG b , PG a , and G4 in old ants, while isopropanol activated five of the focal glomeruli (all except G1) in young ants only ... Additionally, propionic acid activated G4 in young and old ants, while none of the other general odorants activated any of the six focal glomeruli",
     "statistics": [
      {
       "about": "young, PGb, 3-hexanone",
       "test": "one-sided one-sample Welch's t test against zero",
       "p": 4.9e-05
      },
      {
       "about": "young, PGb, isopropanol",
       "test": "one-sided one-sample Welch's t test against zero",
       "p": 0.0063
      },
      {
       "about": "young, PGa, 3-hexanone",
       "test": "one-sided one-sample Welch's t test against zero",
       "p": 0.0025
      },
      {
       "about": "young, PGa, isopropanol",
       "test": "one-sided one-sample Welch's t test against zero",
       "p": 0.0074
      },
      {
       "about": "young, G1, 3-hexanone",
       "test": "one-sided one-sample Welch's t test against zero",
       "p": 0.021
      },
      {
       "about": "young, G2, 3-hexanone",
       "test": "one-sided one-sample Welch's t test against zero",
       "p": 0.00058
      },
      {
       "about": "young, G2, isopropanol",
       "test": "one-sided one-sample Welch's t test against zero",
       "p": 0.0011
      },
      {
       "about": "young, G3, 3-hexanone",
       "test": "one-sided one-sample Welch's t test against zero",
       "p": 0.00011
      },
      {
       "about": "young, G3, isopropanol",
       "test": "one-sided one-sample Welch's t test against zero",
       "p": 0.0036
      },
      {
       "about": "young, G4, 3-hexanone",
       "test": "one-sided one-sample Welch's t test against zero",
       "p": 0.00031
      },
      {
       "about": "young, G4, isopropanol",
       "test": "one-sided one-sample Welch's t test against zero",
       "p": 0.0033
      },
      {
       "about": "young, G4, propionic acid",
       "test": "one-sided one-sample Welch's t test against zero",
       "p": 0.033
      },
      {
       "about": "old, PGb, 3-hexanone",
       "test": "one-sided one-sample Welch's t test against zero",
       "p": 0.00018
      },
      {
       "about": "old, PGa, 3-hexanone",
       "test": "one-sided one-sample Welch's t test against zero",
       "p": 0.011
      },
      {
       "about": "old, G4, 3-hexanone",
       "test": "one-sided one-sample Welch's t test against zero",
       "p": 0.027
      },
      {
       "about": "old, G4, propionic acid",
       "test": "one-sided one-sample Welch's t test against zero",
       "p": 0.023
      }
     ],
     "kind": "finding",
     "inferred": "",
     "outcome": "not_reproduced",
     "blocked_by": null,
     "notebook": "reproduce_hart_2024.ipynb",
     "notes": "At the held-out placement (reading a, radius 8) 56 of 60 cells agree, with no unexpected cell significant. The four that differ are young G1 with 3-hexanone (p = 0.058), old G4 with 3-hexanone (p = 0.15) and G4 with propionic acid at both ages (p = 0.44 and 0.39). The 3-hexanone misses break the text's pattern, so the claim is not reproduced. One of nine variants, the placement from all trials restricted to placements above chance, gives 58 of 60 with the pattern holding (partially reproduced); the placement from all trials gives 57 of 60 and fails on old G4 with 3-hexanone (p = 0.063), as in round 1. Data S2's own values give all 60.",
     "deviations": [
      "Glomerulus positions are not in the dandiset; each is placed where the dandiset's per-trial responses best correlate with Data S2's (data-s2-glomerulus-link), so the test uses Data S2 (data with_inputs).",
      "Assumed the held-out placement as the main measure: each odor's AUCs are read where the glomerulus was placed from Data S2's trials of the other odors (leave one odor out), so that the tested values did not choose the place (verifier, round 1, problem 2). The placement from all trials (round 1's main measure), placements above chance only, and for the alarm pheromones the placement from the general odorants alone are shown as variants.",
      "Assumed a disk ROI of 8 pixels radius (2.3 um) on the three-plane projection; the paper gives no ROI size. Radii 5 and 12 are shown.",
      "Assumed fold-change reading (a), the ROI mean of the projected signal turned into fold change, the usual ImageJ order for \"a max Z-projection was generated ... and the fluorescence fold change was calculated\" (STAR Methods, Image processing); reading (b) is shown, leaving out pixels whose baseline F0 is at or below the ant's background level (the median of the raw maximum projection outside the AL), an assumption, since a pixel at background level divides by noise.",
      "Per ant, the mean over the trials Data S2 holds for that glomerulus (the paper's own trial selection); all trials are shown as a variant.",
      "The 40-frame traces are aligned with one set of shifts per trial, measured on the middle plane, not per plane."
     ],
     "assumed": true,
     "disagreed": false,
     "data": "with_inputs",
     "where": null,
     "proposed": null
    },
    {
     "claim": "fig4-general-odorants-decline",
     "statement": "Responses to 3-hexanone and isopropanol decline significantly with age in five of the six focal glomeruli (3-hexanone in all but PGb, isopropanol in all but G1).",
     "source": "Results, paragraph 9; Figure 4; Figure S4B",
     "quote": "Directly comparing calcium responses across age cohorts showed that responses to 3-hexanone and isopropanol each significantly declined with age in 5 out of 6 glomeruli (3-hexanone, all except PG b ; isopropanol, all except G1)",
     "statistics": [
      {
       "about": "3-hexanone, PGa",
       "test": "Welch's t test",
       "p": 0.014
      },
      {
       "about": "3-hexanone, G1",
       "test": "Welch's t test",
       "p": 0.026
      },
      {
       "about": "3-hexanone, G2",
       "test": "Welch's t test",
       "p": 0.0012
      },
      {
       "about": "3-hexanone, G3",
       "test": "Welch's t test",
       "p": 0.00017
      },
      {
       "about": "3-hexanone, G4",
       "test": "Welch's t test",
       "p": 0.0015
      },
      {
       "about": "isopropanol, PGb",
       "test": "Welch's t test",
       "p": 0.044
      },
      {
       "about": "isopropanol, PGa",
       "test": "Welch's t test",
       "p": 0.011
      },
      {
       "about": "isopropanol, G2",
       "test": "Welch's t test",
       "p": 0.0024
      },
      {
       "about": "isopropanol, G3",
       "test": "Welch's t test",
       "p": 0.008
      },
      {
       "about": "isopropanol, G4",
       "test": "Welch's t test",
       "p": 0.0084
      }
     ],
     "kind": "finding",
     "inferred": "",
     "outcome": "partially_reproduced",
     "blocked_by": null,
     "notebook": "reproduce_hart_2024.ipynb",
     "notes": "At the held-out placement (reading a, radius 8) 9 of the 10 declines are significant: 3-hexanone in PGa, G2, G3 and G4 (p = 0.016, 0.0498, 0.048, 0.020) but not G1 (p = 0.12), isopropanol in all five (PGb 0.008, PGa 0.0098, G2 0.019, G3 0.021, G4 0.025), so at least four of five decline for each odorant, the partial level. 3-hexanone in PGb, not significant in the paper, declines here (29.0 against 12.8, p = 0.022); isopropanol in G1 does not (p = 0.20), as in the paper. The partial level rests on two p-values just under 0.05 and is not met under any of the eight other variants (5 to 8 of 10, one odorant 3 or fewer of 5), including the placement from all trials (7 of 10, isopropanol 3 of 5; not reproduced in round 1). With all trials in place of Data S2's trials, G2 with 3-hexanone gives p = 0.059. Data S2's own values give all 10, with PGb 3-hexanone p = 0.23. The located ROIs bias this claim toward passing. On held-out trials only 9 of 52 old-ant glomerulus traces follow Data S2's, so old-ant values are read partly where the glomerulus is not, which pulls old means down; claims that need old ants to be lower (this one) are pushed toward passing, those that need them higher (fig3d, the late window of figs4d) toward failing. The PGb decline with 3-hexanone is the visible case (old mean 12.8 here against Data S2's 25.1 for the same ants). Changed from not reproduced with the held-out placement.",
     "deviations": [
      "Glomerulus positions are not in the dandiset; each is placed where the dandiset's per-trial responses best correlate with Data S2's (data-s2-glomerulus-link), so the test uses Data S2 (data with_inputs).",
      "Assumed the held-out placement as the main measure: each odor's AUCs are read where the glomerulus was placed from Data S2's trials of the other odors (leave one odor out), so that the tested values did not choose the place (verifier, round 1, problem 2). The placement from all trials (round 1's main measure), placements above chance only, and for the alarm pheromones the placement from the general odorants alone are shown as variants.",
      "Assumed a disk ROI of 8 pixels radius (2.3 um) on the three-plane projection; the paper gives no ROI size. Radii 5 and 12 are shown.",
      "Assumed fold-change reading (a), the ROI mean of the projected signal turned into fold change, the usual ImageJ order for \"a max Z-projection was generated ... and the fluorescence fold change was calculated\" (STAR Methods, Image processing); reading (b) is shown, leaving out pixels whose baseline F0 is at or below the ant's background level (the median of the raw maximum projection outside the AL), an assumption, since a pixel at background level divides by noise.",
      "Per ant, the mean over the trials Data S2 holds for that glomerulus (the paper's own trial selection); all trials are shown as a variant.",
      "The 40-frame traces are aligned with one set of shifts per trial, measured on the middle plane, not per plane."
     ],
     "assumed": true,
     "disagreed": false,
     "data": "with_inputs",
     "where": null,
     "proposed": null
    },
    {
     "claim": "fig4-propionic-acid-g2-g3",
     "statement": "In young ants propionic acid gives a very small negative deflection in G2 and G3 that declines in amplitude with age; its responses in G4 tend to fall with age, not significantly.",
     "source": "Results, paragraph 9; Figure 4; Figure S4B",
     "quote": "We also found that, in young ants, propionic acid produced a very small negative deflection in G2 and G3 that declined in amplitude with age ... Propionic acid responses in G4 also tended to decrease with age, but the difference was not statistically significant",
     "statistics": [
      {
       "about": "G2, young against old",
       "test": "Welch's t test",
       "p": 0.0052
      },
      {
       "about": "G3, young against old",
       "test": "Welch's t test",
       "p": 0.029
      },
      {
       "about": "G4, young against old",
       "test": "Welch's t test",
       "p": 0.05,
       "p_relation": ">"
      }
     ],
     "kind": "finding",
     "inferred": "",
     "outcome": "not_reproduced",
     "blocked_by": null,
     "notebook": "reproduce_hart_2024.ipynb",
     "notes": "At the held-out placement (reading a, radius 8) the young propionic acid AUC is negative in G2 (-5.1) and G3 (-4.9) and nearer zero in old ants (-3.3 and -0.83), the stated direction, but neither difference is significant (p = 0.60 and 0.17); no variant meets the criterion (round 1's placement p = 0.28 and 0.12). G4 does not differ (0.92 against 0.75, p = 0.98). Data S2's own values give G2 p = 0.0050 and G3 p = 0.030.",
     "deviations": [
      "Glomerulus positions are not in the dandiset; each is placed where the dandiset's per-trial responses best correlate with Data S2's (data-s2-glomerulus-link), so the test uses Data S2 (data with_inputs).",
      "Assumed the held-out placement as the main measure: each odor's AUCs are read where the glomerulus was placed from Data S2's trials of the other odors (leave one odor out), so that the tested values did not choose the place (verifier, round 1, problem 2). The placement from all trials (round 1's main measure), placements above chance only, and for the alarm pheromones the placement from the general odorants alone are shown as variants.",
      "Assumed a disk ROI of 8 pixels radius (2.3 um) on the three-plane projection; the paper gives no ROI size. Radii 5 and 12 are shown.",
      "Assumed fold-change reading (a), the ROI mean of the projected signal turned into fold change, the usual ImageJ order for \"a max Z-projection was generated ... and the fluorescence fold change was calculated\" (STAR Methods, Image processing); reading (b) is shown, leaving out pixels whose baseline F0 is at or below the ant's background level (the median of the raw maximum projection outside the AL), an assumption, since a pixel at background level divides by noise.",
      "Per ant, the mean over the trials Data S2 holds for that glomerulus (the paper's own trial selection); all trials are shown as a variant.",
      "The 40-frame traces are aligned with one set of shifts per trial, measured on the middle plane, not per plane."
     ],
     "assumed": true,
     "disagreed": false,
     "data": "with_inputs",
     "where": null,
     "proposed": null
    },
    {
     "claim": "figs4d-pgb-hexanone-dynamics",
     "statement": "In PGb, 3-hexanone responses of old ants have a lower peak but decline more slowly than those of young ants; the AUC over frames 9 to 14 is larger in young ants and over frames 35 to 40 larger in old ants.",
     "source": "Results, paragraph 10; Figure S4C and S4D",
     "quote": "responses to 3-hexanone in PG b had a lower peak amplitude but then declined more slowly in old versus young ants",
     "statistics": [
      {
       "about": "frames 9 to 14, young against old",
       "test": "Welch's t test",
       "p": 0.002,
       "n": "7 young and 7 old ants"
      },
      {
       "about": "frames 35 to 40, young against old",
       "test": "Welch's t test",
       "p": 0.0098,
       "n": "7 young and 7 old ants"
      }
     ],
     "kind": "finding",
     "inferred": "",
     "outcome": "partially_reproduced",
     "blocked_by": null,
     "notebook": "reproduce_hart_2024.ipynb",
     "notes": "At the held-out placement (reading a, radius 8) the PGb response to 3-hexanone over frames 9 to 14 is larger in young ants (9.5 against 3.5, p = 0.0026; every variant p <= 0.0061). Over frames 35 to 40 old ants are not above young ones (0.68 against 0.98, p = 0.67; variants 0.40 to 0.96). Data S2's own values give both (p = 0.0020 and 0.0098).",
     "deviations": [
      "Glomerulus positions are not in the dandiset; each is placed where the dandiset's per-trial responses best correlate with Data S2's (data-s2-glomerulus-link), so the test uses Data S2 (data with_inputs).",
      "Assumed the held-out placement as the main measure: each odor's AUCs are read where the glomerulus was placed from Data S2's trials of the other odors (leave one odor out), so that the tested values did not choose the place (verifier, round 1, problem 2). The placement from all trials (round 1's main measure), placements above chance only, and for the alarm pheromones the placement from the general odorants alone are shown as variants.",
      "Assumed a disk ROI of 8 pixels radius (2.3 um) on the three-plane projection; the paper gives no ROI size. Radii 5 and 12 are shown.",
      "Assumed fold-change reading (a), the ROI mean of the projected signal turned into fold change, the usual ImageJ order for \"a max Z-projection was generated ... and the fluorescence fold change was calculated\" (STAR Methods, Image processing); reading (b) is shown, leaving out pixels whose baseline F0 is at or below the ant's background level (the median of the raw maximum projection outside the AL), an assumption, since a pixel at background level divides by noise.",
      "Per ant, the mean over the trials Data S2 holds for that glomerulus (the paper's own trial selection); all trials are shown as a variant.",
      "The 40-frame traces are aligned with one set of shifts per trial, measured on the middle plane, not per plane."
     ],
     "assumed": true,
     "disagreed": false,
     "data": "with_inputs",
     "where": null,
     "proposed": null
    },
    {
     "claim": "pg-opposite-modulation",
     "statement": "In PGb and PGa, responses to alarm pheromones rise with age while responses to general odorants fall or stay absent; no general odorant response rises with age in any focal glomerulus.",
     "source": "Results, paragraph 9; Figures 3, 4 and S4B",
     "quote": "Whereas alarm pheromone responses increased with age specifically in PG b and PG a , general odorant responses showed overall similar patterns across all six glomeruli, either with no responses or decreasing responses with age",
     "statistics": [],
     "kind": "finding",
     "inferred": "",
     "outcome": "not_reproduced",
     "blocked_by": null,
     "notebook": "reproduce_hart_2024.ipynb",
     "notes": "The criterion needs fig3d-pg-stronger-old, which is not reproduced, so the claim is not reproduced. The general odorant parts hold: no general odorant response rises with age in any focal glomerulus (0 of 30 Welch tests under every variant), and PGb and PGa responses to 3-hexanone and isopropanol fall with age (4 of 4 significant at the held-out placement). The criterion states only when the claim is reproduced; round 1 gave partially reproduced, changed after the round 1 verifier. On Data S2's own values 2 of the 30 tests have old significantly higher (G2 and G3 with propionic acid, p = 0.0050 and 0.030), the change that fig4-propionic-acid-g2-g3 expects, so the two criteria cannot both hold on the paper's data; this is asked of a reader (questions/000773.yaml).",
     "deviations": [
      "Glomerulus positions are not in the dandiset; each is placed where the dandiset's per-trial responses best correlate with Data S2's (data-s2-glomerulus-link), so the test uses Data S2 (data with_inputs).",
      "Assumed the held-out placement as the main measure: each odor's AUCs are read where the glomerulus was placed from Data S2's trials of the other odors (leave one odor out), so that the tested values did not choose the place (verifier, round 1, problem 2). The placement from all trials (round 1's main measure), placements above chance only, and for the alarm pheromones the placement from the general odorants alone are shown as variants.",
      "Assumed a disk ROI of 8 pixels radius (2.3 um) on the three-plane projection; the paper gives no ROI size. Radii 5 and 12 are shown.",
      "Assumed fold-change reading (a), the ROI mean of the projected signal turned into fold change, the usual ImageJ order for \"a max Z-projection was generated ... and the fluorescence fold change was calculated\" (STAR Methods, Image processing); reading (b) is shown, leaving out pixels whose baseline F0 is at or below the ant's background level (the median of the raw maximum projection outside the AL), an assumption, since a pixel at background level divides by noise.",
      "Per ant, the mean over the trials Data S2 holds for that glomerulus (the paper's own trial selection); all trials are shown as a variant.",
      "The 40-frame traces are aligned with one set of shifts per trial, measured on the middle plane, not per plane."
     ],
     "assumed": true,
     "disagreed": true,
     "data": "with_inputs",
     "where": null,
     "proposed": null
    }
   ],
   "findings": [
    {
     "kind": "metadata_insufficient",
     "statement": "The dandiset holds no ROI positions, glomerulus identities or responding-ROI counts. The six focal glomeruli (PGb, PGa, G1 to G4) and the responding ROIs of Figures 2E and S3C were drawn by hand and cannot be located from the dandiset alone; twelve claims depend on them (eight list the link claim as a dependency, pg-opposite-modulation depends on it through others, and three rest on the responding ROIs). Data S2 of the paper holds the traces of the focal glomeruli but not their positions.",
     "found": null,
     "status": null
    },
    {
     "kind": "metadata_insufficient",
     "statement": "Stimulus timing is not stored as times or intervals, only in the series description (\"odor stimulus triggered at 3s with duration 5s\"). The traces are consistent with it; the response first exceeds baseline at frame 6 and peaks at frame 12, and paraffin oil trials show no response.",
     "found": null,
     "status": null
    },
    {
     "kind": "metadata_error",
     "statement": "The timing metadata disagree. The series description gives 40 cycles in 48.0 s (1.2 s per volume, as 33 planes at 27.5 frames per second take), the series rate is 1.2097 Hz (0.83 s per volume, the Methods' value), and the imaging plane's imaging_rate is 0.83, a period stored as a rate.",
     "found": null,
     "status": null
    },
    {
     "kind": "metadata_error",
     "statement": "One trial (ant1-m2-d24-y2023, trial 135) spells ethylpyrazine \"ethylyprazine\" in its 33 file names; its place in the odor sequence shows it is ethylpyrazine.",
     "found": null,
     "status": null
    },
    {
     "kind": "selection_undocumented",
     "statement": "The Methods say a small number of trials were excluded from the responding-ROI counts for motion artifacts, but neither the dandiset nor Data S2 says which; Data S2 holds all 21 odor trials of every 48% ant.",
     "found": null,
     "status": null
    },
    {
     "kind": "methods_unavailable",
     "statement": "Question for the paper's authors. How were the responding ROIs of Figure 2E and S3C drawn (their size, and their positions per ant), and what are the per-ant counts? With the claims file's automated ROIs the medians for 4-methyl-3-heptanone are 4 and 2 (the paper's values) at 7 um glomeruli, 1 and 0 at 10 um and 0 and 0 at 14 um, and no size gives a significant age difference (p 0.22 to 0.71).",
     "found": null,
     "status": null
    },
    {
     "kind": "methods_unavailable",
     "statement": "Question for the paper's authors. Where are the six focal glomeruli in each ant (plane and position), what size were their ROIs, and was the fold change computed on the ROI mean or per pixel? Were the PGb and PGa alarm traces of Figure 3C and 3D (Data S2 names with a two-digit year) quantified on the same ROIs and planes, and with the same processing, as the general odorant traces of Figure 4? At positions matched to Data S2, PGb alarm AUCs measured from the imaging are well below Data S2's in the 60-day ants of this dandiset (5.15 against 20.66 for 4-methyl-3-heptanone) and in the 90-day ants of DANDI 000467 (11.59 against 24.99), although the trace shapes often match. Placed by correlation with Data S2, every young-ant glomerulus is above chance but 15 of 39 old-ant glomeruli are not.",
     "found": null,
     "status": null
    },
    {
     "kind": "other",
     "statement": "The whole-AL mean of the peak fold change map is negative for a non-responding AL (the paraffin oil trials have a median of -0.23), because the minimum filter is applied to noisy per-pixel fold change before the maximum projection; without the filter the map saturates near 4. The fold declines of Figure S3B (more than 9-fold and 57-fold) are therefore sensitive to the baseline; here the old cohort's isopropanol mean is negative, and with each ant's paraffin oil value subtracted the ratios are 6.0 and 9.5.",
     "found": null,
     "status": null
    },
    {
     "kind": "other",
     "statement": "The behaviour data of Figure 1 and S1 are only in the paper's supplement (Data S1), the confocal stacks of Figure 2A and 2B only in the Brain Image Library (doi:10.35077/g.1153), and the 90-day ants of Figure 3C and 3D in DANDI 000467. The Brain Image Library download server could not be reached during this attempt. In Data S2 the trailing number of a trial name is the repeat for most ants but the recording's trial number for the 90-day PGb traces; one of those 96 numbers does not match the 000467 trial of that odor and repeat.",
     "found": null,
     "status": null
    },
    {
     "kind": "other",
     "statement": "Raw fluorescence is stored as uncompressed float64 (8 bytes per value), 1.28 TB for 21 ants; each file is one plane of one trial (15,246 files). Reading needs only range requests, but the dataset is large for what it holds.",
     "found": null,
     "status": null
    }
   ],
   "reconciliation": null,
   "verifications": [
    {
     "date": "2026-10-31",
     "kind": "agent",
     "by": "Independent agent run by CatalystNeuro (same harness, separate context, did not write the notebook)",
     "reviewed": "The handover of the first run (the executed notebook with the conclusion cells and the summary removed, the helper module antal.py and the claims file), with the paper under paper/; the verifier also listed the dandiset's files and metadata for the blocked claims. Report: transcripts/000773/verifier-1-report.md. Problems 10 and 11, which do not affect a conclusion, were answered in the entry: the roi_trace docstring now says the ROI follows the per-trial shifts, and the G4 alarm n of 5 old ants is because Data S2 holds only general odorant trials for G4 of ant4-m1-d10 (the all-trials variant has 6).",
     "agreed": 19,
     "of": 22,
     "disagreements": [
      {
       "claim": "figs3c-general-breadth-no-age-effect",
       "outcome": "blocked",
       "reason": "Blocked (methods_underspecified). By its letter all three p >= 0.05, but it uses the same automated counts the notebook declined to trust for fig2e, and that procedure misses the paper's positive effect; a null from it is not a test. Problem 7: those counts are implausible (up to 173 ROIs per ant; 72 ROIs responding to ethanol in ant4-m1-d20) and change with diameter."
      },
      {
       "claim": "fig3d-pg-60-vs-90-days-same",
       "outcome": "blocked",
       "reason": "Blocked (methods_underspecified). Reading (a): all p >= 0.15. Reading (b): PGb with 4-methyl-3-heptanone p = 0.0002. Problem 3: the 90-day PGa positions are found from 6 trials, which the notebook itself says should not be used; their r of 0.996-0.9997 is at chance level and 3 of 8 fail the AUC check. Problem 4: reading (b) compares 60-day AUCs from the dandiset with 90-day AUCs from Data S2, two pipelines (about 4 times apart for the same 60-day ants); the p = 0.0002 reflects that scale, not age, and this mixing is built into the claims file's reading (b)."
      },
      {
       "claim": "pg-opposite-modulation",
       "outcome": "not_reproduced",
       "reason": "This claim depends on fig3d-pg-stronger-old, which fails. The other parts hold: 0 of 30 general-odorant tests have old significantly higher, and 2 PGb/PGa declines are significant. Problem 8: the criterion requires none of the 30 tests to have old significantly higher, but fig4-propionic-acid-g2-g3 expects old propionic-acid responses in G2 and G3 significantly nearer zero, which is higher; on Data S2's own values this condition fails (2 of 30). The two criteria contradict each other."
      }
     ]
    },
    {
     "date": "2026-10-31",
     "kind": "agent",
     "by": "Independent agent run by CatalystNeuro (same harness, separate context, did not write the notebook)",
     "reviewed": "The full executed notebook of the first revision (79 cells, with the conclusion cells and the summary), the helper module antal.py, the claims file, the attempt record and the README, with the paper under paper/. Report: transcripts/000773/verifier-2-report.md. The verifier agreed with all 22 outcomes. Points answered in the entry and this record: the fig3b count of old ants (5, not 4); the fig2b timeouts, now as printed, with the landing page read and its notices printed by the notebook, and the shell attempts said to be in no output; the README and summary sentences on Data S2, now excepting pg-opposite-modulation; the summary's \"need no hand-drawn ROI\", now saying the outline is a threshold stand-in; the run time (36.2 minutes in this run, 43.9 in round 2), now about 40 minutes; the number of claims that depend on the hand-drawn ROIs, now twelve; and the figs2a smallest p, now 0.0799. Added: the direction of the placement bias in the fig4 notes and the summary, the figs3c caveat in the figs2a notes, and the held-out 60-day values against the 000467 90-day values for fig3d-pg-60-vs-90-days-same. The claims-file problems the verifier lists are in questions/000773.yaml; the missing rule for readings that disagree in fig3d-pg-60-vs-90-days-same is added there.",
     "agreed": 22,
     "of": 22,
     "disagreements": []
    }
   ]
  }
 ],
 "entries": [
  {
   "name": "000773-hart-2024",
   "type": "reproduction",
   "title": "Age and odor coding in the clonal raider ant antennal lobe (Hart et al. 2024)",
   "description": "Tests the findings of Hart, Lopes, Frank and Kronauer (2024) on dandiset 000773, which holds the two-photon calcium imaging of the right antennal lobe of 21 young and old GCaMP6s ants stimulated with two alarm pheromones and five general odorants. It computes the whole-lobe peak response maps, counts responding ROIs, and measures the six focal glomeruli, located with the paper's supplementary traces (Data S2). The behaviour claims are tested on the paper's Data S1.",
   "datasets": [
    "dandi:000773"
   ],
   "papers": [
    "10.1016/j.cub.2024.05.031"
   ],
   "derived_from": [],
   "notebooks": [
    {
     "path": "reproduce_hart_2024.ipynb",
     "title": "Reproducing Hart et al. (2024) from dandiset 000773",
     "executed": "notebooks/000773-hart-2024/reproduce_hart_2024.html"
    }
   ]
  }
 ]
}