{ "dataset": { "archive": "dandi", "id": "000166", "label": "DANDI:000166" }, "page": "datasets/dandi/000166/", "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": 9, "partially_reproduced": 9, "not_reproduced": 23, "blocked": 5 }, "outcome_counts_with_outside_data": { "reproduced": 0, "partially_reproduced": 0, "not_reproduced": 0, "blocked": 0 }, "outcome_counts_elsewhere": { "reproduced": 0, "partially_reproduced": 0, "not_reproduced": 0, "blocked": 1 }, "outcome_counts_links": { "reproduced": 0, "partially_reproduced": 0, "not_reproduced": 0, "blocked": 0 }, "attempts": [ { "id": "000166-senzai-2019-01", "dataset": { "archive": "dandi", "id": "000166", "version": "0.250624.0434", "doi": "10.48324/dandi.000166/0.250624.0434" }, "key": "dandi:000166", "label": "DANDI:000166", "dataset_version": "0.250624.0434", "paper": "10.1016/j.neuron.2018.12.009", "paper_title": "Publication", "date": "2026-10-29", "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/40815140604" }, "tier": "agent_verified", "deprecated": "", "deprecated_on": null, "entry": "000166-senzai-2019", "entry_title": "Layer-specific features of mouse V1 units (Senzai et al. 2019)", "description": "Tests the findings of Senzai, Fernandez-Ruiz and Buzsaki (2019) on dandiset 000166, which holds one session from each of 19 mice recorded with a 64-site linear probe in V1, with spike-sorted units, LFP, the raw signal, brain state and UP/DOWN intervals and laser pulse times. It tests the unit counts and waveform classes, the CCG-based detection of monosynaptic connections, the optogenetic tagging, the gamma ICA and coherence clusters, the laminar LFP power, cross-frequency coupling and 3-6 Hz phase locking, the K-means unit clusters and the layer features of firing and connections that rest on them, and the DOWN state-active neurons. Depth is aligned on the MUA power peak, as the CSD landmarks cannot be placed by a rule the paper gives.", "notebooks": [ { "path": "reproduce_senzai_2019.ipynb", "title": "Reproducing Senzai et al. (2019) from dandiset 000166", "executed": "notebooks/000166-senzai-2019/reproduce_senzai_2019.html" } ], "outcomes": [ { "claim": "fig1f-sink-b-deeper", "statement": "The layer 3/4 border sink of the DOWN-UP transition of non-REM slow oscillations (landmark b) lies slightly but significantly deeper than the early light-evoked sink (b') in the same mice, by 39 \u00b1 49 um.", "source": "Results, Anchoring Mesoscopic Patterns to Physiological Landmarks, paragraph 2; Figure 1F", "quote": "Sink b, associated with the \u201cUP\u201d state of slow oscillation during non-REM sleep, was slightly but significantly deeper than the light-evoked sink b\u2032 in the layer 3/4 border (39 \u00b1 49 \u03bcm", "statistics": [ { "test": "signed rank test", "p": 0.05, "p_relation": "<", "n": "13 mice" } ], "kind": "finding", "inferred": "", "outcome": "blocked", "blocked_by": "data_missing", "notebook": "reproduce_senzai_2019.ipynb", "notes": "The light-evoked CSD needs the times of the 520 nm flashes in the 13 stimulated mice. The dataset holds them in 2 files only, the `Visual laser` tables of YMV05 and YMV08 (498 intervals of 0.4 s each, plus the optogenetic pulses). With 2 mice a signed-rank test cannot reach p < 0.05 (its smallest two-sided p is 0.5). The DOWN-UP half of the comparison is computed under fig1g-landmark-distances-consistent.", "deviations": [], "assumed": false, "disagreed": false, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig1f-other-landmarks-same-depth", "statement": "The other CSD sinks and sources of slow oscillations (a, d, e) occur at the same depths as the corresponding light-evoked sinks and sources (a', d', e').", "source": "Results, Anchoring Mesoscopic Patterns to Physiological Landmarks, paragraph 2; Figure 1F", "quote": "whereas the other sinks and sources occurred at the same depth as those seen with visual stimulation (p > 0.05, signed rank test)", "statistics": [ { "test": "signed rank test", "p": 0.05, "p_relation": ">", "n": "13 mice" } ], "kind": "finding", "inferred": "", "outcome": "blocked", "blocked_by": "data_missing", "notebook": "reproduce_senzai_2019.ipynb", "notes": "Needs the light-evoked landmarks a', d' and e', which need the flash times of 13 mice; the dataset has them for 2 (see fig1f-sink-b-deeper). It also depends on fig1f-sink-b-deeper.", "deviations": [], "assumed": false, "disagreed": false, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig1g-landmark-distances-consistent", "statement": "The distances between the five depth landmarks (the MUA power peak c and the CSD landmarks a, b, d, e) vary by less than 100 um from mouse to mouse.", "source": "Results, Anchoring Mesoscopic Patterns to Physiological Landmarks, paragraph 2; Figure 1G", "quote": "Importantly, the distances between the five landmarks were consistent across animals and varied <100 \u03bcm from mouse to mouse ( Figure 1 G; n = 19)", "statistics": [], "kind": "finding", "inferred": "", "outcome": "blocked", "blocked_by": "methods_underspecified", "notebook": "reproduce_senzai_2019.ipynb", "notes": "Every reading was run (CSD as the unsmoothed second difference of the claims file and, as a variant, over two sites; windows 0-100 and 0-50 ms; \"varied\" as the range and as the largest deviation from the mean). The data columns are in depth order (for each file, the best-correlated site in the 100-500 Hz LFP is an adjacent one for 50 to 63 of the 64 sites, median 58). The MUA peak c is found in all 19 mice, but the rule (most negative CSD for b above c and d below it, most positive for a above b and e below d, each over the whole half of the probe) places all five landmarks in only 15 or 16 mice, and among those the ranges of the adjacent distances are 220 to 620 um (largest deviations 114 to 420 um), against less than 110 um. The CSD maps of all mice show that the rule picks probe-edge values in some mice (a or b within 3 sites of the top in YMV09 and YMV13, d or e within 3 sites of the bottom in YMV09) and places a one site above b in 7 mice. The paper places the landmarks by eye (\"the most prominent current source density (CSD) sinks and sources\", Results, Anchoring Mesoscopic Patterns, paragraph 1; asterisks in Figure 1D), and neither it nor the claims file gives a rule that can be run, so the claim is blocked with every reading shown. The question for the paper's authors is in the findings. The other laminar claims are tested with depth relative to c, which needs no other landmark.", "deviations": [ "DOWN-UP transitions are the dataset's UP onsets that follow a DOWN interval, averaged over the 1000 s of each session holding the most of them (114 to 1,541 transitions), not over the whole session.", "MUA power (landmark c) is computed from the two 30 s raw blocks used for the waveforms, not the whole session.", "Site 0 is taken as the most superficial, as the files' rel_y runs from -1 to -64; the electrode table gives no depth." ], "assumed": false, "disagreed": false, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig2b-six-coherence-clusters", "statement": "Clustering the 64 sites by their pairwise gamma (30-100 Hz) coherence with a gradient-descent algorithm gives six clusters whose five boundaries are consistent across mice in normalized depth.", "source": "Results, Anchoring Mesoscopic Patterns to Physiological Landmarks, paragraph 3; Figures 2A, 2B", "quote": "the recording sites were clustered on the basis of their resulting cross-coherence matrix, resulting in six clusters whose five boundaries were consistent across animals when projected into normalized depth coordinates", "statistics": [ { "test": "signed rank test for cluster boundaries across animals", "p": 0.05, "p_relation": ">", "n": "19 mice" } ], "kind": "finding", "inferred": "", "outcome": "reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "Under the readings the criterion lists with six clusters imposed (reading b) and each mouse's boundaries compared with the across-mouse median, the claim holds: a six-cluster solution is found in all 18 mice (contiguous in depth in 14) and no boundary differs from the median (signed-rank p 0.74 to 1). The other readings fail: left free, the clustering converges to 8 to 20 clusters from 64 random labels (6 to 10 from 16), and against the matching IC boundaries three of five boundaries differ (p = 0.033, 0.013, 0.028; median differences 3.5 to 5.8 sites). The comparison with the across-mouse median can hardly fail by construction, so the outcome shows only that six clusters can be placed with boundaries spread around common depths (medians 21, 13.5 and 3 sites above c and 4.5 and 12.5 below it). Each boundary ranges across mice by 31 to 40 sites (620 to 800 um; boundary 1 from 46.5 to 12.5 sites above c), wider than the 220 to 620 um spread of the landmark distances that fig1g-landmark-distances-consistent calls inconsistent. The criterion's reading (b) for the cluster number together with its reading (b) for the boundaries can hardly fail; this is asked of a reader in questions/000166.yaml.", "deviations": [ "Coherence is the magnitude-squared coherence in 1 s Hann segments, averaged over 30-100 Hz, on the waking blocks of the ICA (up to 600 s per mouse).", "The energy-gap moves follow the STAR Methods (Gamma coherence and gradient descent clustering). The number of random initial clusters is not given; 64 (and 16, shown) are used for reading (a), and 6 (or 8 when no run of 20 ends with six) for reading (b); the lowest-energy run of 20 is kept.", "Boundary k is the number of sites in clusters 1 to k minus 0.5, exact for contiguous clusters.", "Depth is taken in sites relative to landmark c (the MUA power peak), in place of the paper's normalized depth, which warps depth between all five landmarks (STAR Methods, Depth normalization) and so needs a, b, d and e, which no rule given in the paper places (fig1g-landmark-distances-consistent). Aligning on c is an assumption.", "YMV19 has no state table and is left out." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig2b-clusters-same-across-states", "statement": "The depth distribution of the six gamma coherence clusters corresponds roughly to the six cortical layers in each brain state and does not differ between states.", "source": "Results, Anchoring Mesoscopic Patterns to Physiological Landmarks, paragraph 3; Figure 2B; Figure S2E", "quote": "The depth distribution of the six gamma coherence clusters corresponded roughly to the six cortical layers in each brain state ( Figures 2 B and S2 ; p > 0.05, ANOVA for comparison across different states)", "statistics": [ { "test": "ANOVA for comparison across different states", "p": 0.05, "p_relation": ">", "n": "19 mice" } ], "kind": "finding", "inferred": "", "outcome": "reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "No boundary of the six-cluster solutions differs across waking, non-REM and REM (one-way ANOVA p 0.53, 0.58, 0.71, 0.86, 0.87; Kruskal-Wallis p 0.41 to 0.87; 18 mice per state); the boundary means differ by at most 3.2 sites between states (boundary 1, 23.4 against 20.2 sites above c).", "deviations": [ "Non-REM and REM coherence come from up to 600 s and 300 s of each state; six clusters imposed as in fig2b reading (b).", "Depth is taken in sites relative to landmark c (the MUA power peak), in place of the paper's normalized depth, which warps depth between all five landmarks (STAR Methods, Depth normalization) and so needs a, b, d and e, which no rule given in the paper places (fig1g-landmark-distances-consistent). Aligning on c is an assumption.", "YMV19 has no state table and is left out." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig2c-six-gamma-ics", "statement": "ICA of the 30-100 Hz LFP yields six main independent components whose spatial loadings are highly consistent across mice and whose depth distribution is related to the anatomical layers.", "source": "Results, Anchoring Mesoscopic Patterns to Physiological Landmarks, paragraph 3; Figure 2C", "quote": "ICA yielded six main independent components (ICs) that were highly consistent across animals (p < 0.001, Spearman\u2019s correlation for ICs spatial loadings)", "statistics": [ { "test": "Spearman's correlation for ICs spatial loadings", "p": 0.001, "p_relation": "<", "n": "19 mice" } ], "kind": "finding", "inferred": "", "outcome": "partially_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "With the six highest-variance ICs (reading a), each mouse's loading correlates positively with the mean of the other mice for all six ICs (median Spearman rho 0.87, 0.84, 0.75, 0.58, 0.58, 0.94 from the surface), with signed-rank p below 0.001 for five (p 7.6e-6 to 0.00033) and p = 0.0019 for IC 4, so the criterion's p < 0.001 for each of the six fails for one. The pairwise reading gives p below 0.001 for five ICs and 0.02 for IC 4. With the six confined ICs (reading b), ICs 1-3 pass and ICs 4-6 do not (p 0.0056 to 0.016). The ICs lie in depth from a median 26 sites above c to 24 below it. In 17 of 18 mice one of the six peaks at or near the deepest site and carries 0.03 to 0.71 of the band variance. This deepest IC is main IC 6 under reading (a) and is spread over depth in 9 of 18 mice, so there it is also counted as volume conducted in ica-volume-conduction-fraction reading (a); its loading is the most consistent IC across mice (median rho 0.94). Reading (b) leaves the spread ICs out and fails for ICs 4-6. 18 mice (YMV19 has no state table).", "deviations": [ "The gamma ICA (STAR Methods, Independent Component Analysis of LFPs) is fitted on up to 600 s of waking per mouse (the longest `Awake` epochs of at least 30 s), with scikit-learn's FastICA on the first ten PCs of the 30-100 Hz LFP in place of runica (EEGLAB infomax, not available in Python); the whole-session reading of fig2c is not run.", "Depth is taken in sites relative to landmark c (the MUA power peak), in place of the paper's normalized depth, which warps depth between all five landmarks (STAR Methods, Depth normalization) and so needs a, b, d and e, which no rule given in the paper places (fig1g-landmark-distances-consistent). Aligning on c is an assumption.", "Each mouse's loading is compared with the mean of the other mice (leave-one-out), over aligned sites held by at least 5 other mice, and the p is a signed-rank test of the 18 correlations against zero; the claims file names the comparison but not the test of the correlations.", "A loading is confined when fewer than half the sites exceed half its peak absolute value (a choice, stated; used for reading b).", "YMV19 has no state table and is left out." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "ica-volume-conduction-fraction", "statement": "ICA removes volume-conducted currents, which account for 0.34 \u00b1 0.11 of the V1 gamma LFP variance.", "source": "Results, Anchoring Mesoscopic Patterns to Physiological Landmarks, paragraph 3", "quote": "This method has the additional advantage of removing volume-conducted currents ( Fern\u00e1ndez-Ruiz et al., 2012 ) that accounted for 0.34 \u00b1 0.11 of V1 gamma LFP variance", "statistics": [], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "Reading (a), the ICs whose loadings are spread over depth (half the sites or more above half their peak) carry 0.534 \u00b1 0.163 of the gamma variance (1 to 4 such ICs per mouse); reading (b), one minus the share of the six main ICs, is 0.056 \u00b1 0.022. Neither is within 0.34 \u00b1 0.11. The ten PCs hold 0.99 of the band-passed variance, so reading (b) is bound to be small. In 9 of 18 mice the main IC 6 of fig2c reading (a) is one of the spread ICs, so reading (a) here and fig2c reading (a) count the same component in two roles; under the consistent pairing (spread ICs as volume conducted, confined ICs as main), fig2c reading (b) fails for ICs 4-6 and fig2d still fails.", "deviations": [ "The gamma ICA (STAR Methods, Independent Component Analysis of LFPs) is fitted on up to 600 s of waking per mouse (the longest `Awake` epochs of at least 30 s), with scikit-learn's FastICA on the first ten PCs of the 30-100 Hz LFP in place of runica (EEGLAB infomax, not available in Python); the whole-session reading of fig2c is not run.", "The spread rule (half the sites or more above half the peak loading) is a choice; the claims file reads spread as uniform over depth (Fernandez-Ruiz et al. 2012) without a threshold.", "YMV19 has no state table and is left out." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig2d-superficial-ic-power", "statement": "The three superficial gamma ICs have larger relative power than the three deep ones (0.14 \u00b1 0.04 against 0.08 \u00b1 0.03).", "source": "Results, Anchoring Mesoscopic Patterns to Physiological Landmarks, paragraph 3; Figure 2D", "quote": "The superficial gamma ICs had a significantly larger power compared with the deeper ones (top three versus bottom three; Figures 2 D and S4 ; 0.14 \u00b1 0.04 versus 0.08 \u00b1 0.03 relative power for ICs 1\u20133 versus 4\u20136", "statistics": [ { "test": "rank sum test", "p": 1.1e-07 } ], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "Relative power of ICs 1-3 is 0.106 \u00b1 0.101 (within 0.14 \u00b1 0.04) and of ICs 4-6 0.209 \u00b1 0.200 (outside 0.08 \u00b1 0.03), 54 values each; the deep ICs are larger (rank-sum p = 0.035), the opposite of the claim. The deepest IC carries most of this (median 0.30); it is spread over depth in 9 of 18 mice. With the six confined ICs, which leave the spread ICs out, ICs 1-3 average 0.086 \u00b1 0.079 (outside 0.14 \u00b1 0.04) and ICs 4-6 0.061 \u00b1 0.067 (within 0.08 \u00b1 0.03), superficial larger with rank-sum p = 0.0089, so the claim fails under that reading too.", "deviations": [ "The gamma ICA (STAR Methods, Independent Component Analysis of LFPs) is fitted on up to 600 s of waking per mouse (the longest `Awake` epochs of at least 30 s), with scikit-learn's FastICA on the first ten PCs of the 30-100 Hz LFP in place of runica (EEGLAB infomax, not available in Python); the whole-session reading of fig2c is not run.", "ICs are numbered by the depth of their peak loading; the six are the highest-variance ones (fig2c reading a).", "YMV19 has no state table and is left out." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig3c-slow-oscillation-nrem-deep", "statement": "Slow oscillation power (0.5-3 Hz) is dominant in non-REM sleep and largest in the deeper layers.", "source": "Results, Brain State-Dependent Features of LFPs, paragraph 1; Figures 3C, S3D", "quote": "Slow oscillation power (0.5\u20133 Hz) was dominant in non-REM sleep ... especially in deeper layers", "statistics": [ { "about": "across brain states", "test": "Kruskal-Wallis test", "p": 0.0004, "n": "19 mice" }, { "about": "across layers", "test": "Kruskal-Wallis test", "p": 4.1e-05 } ], "kind": "finding", "inferred": "", "outcome": "partially_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "The state part holds: 0.5-3 Hz power (mean over sites) is highest in non-REM (median over mice 6,594 against 5,268 in waking and 1,514 in REM uV^2/Hz; Kruskal-Wallis p = 5.3e-5). Power differs across the six layers in non-REM (p = 9.8e-6 with coherence layers, 2.5e-6 with IC layers), but the largest is in the fourth layer under both readings (normalized medians 1.47 and 1.44; fifth 1.42 and 1.43; sixth 0.95 and 0.80), not in the deepest two.", "deviations": [ "Power is the Welch spectrum (4096-sample segments) of up to 600 s of each state, averaged over the band, a standard choice equal in its band mean to a wavelet power average.", "Coherence layers are the waking six-cluster solution of fig2b (reading b); IC layers assign each site to the nearest peak of the six main ICs.", "YMV19 has no state table and is left out." ], "assumed": false, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig3d-slow-oscillation-gamma-coupling", "statement": "In non-REM sleep the slow oscillation phase modulates gamma amplitude in superficial layers (69 \u00b1 12 Hz, modulation index 2.7 \u00b1 0.5 \u00d7 10^-3) and high gamma (>100 Hz) in deep layers (modulation index 2.3 \u00b1 0.3 \u00d7 10^-3).", "source": "Results, Brain State-Dependent Features of LFPs, paragraph 1; Figure 3D, left", "quote": "Slow oscillation was phase-coupled to gamma frequency power in superficial layers (69 \u00b1 12 Hz; modulation index 2.7 \u00b1 0.5 \u00d7 10 \u22123 ) and to the high-gamma (>100 Hz) band in deep layers", "statistics": [], "kind": "finding", "inferred": "", "outcome": "partially_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "With the 1-3 Hz phase of the layer 6 site in non-REM (reading a), the index averaged over the sites above the layer 4 site peaks at 70.9 \u00b1 16.2 Hz over 18 mice (69 \u00b1 12), within tolerance. The indices are larger than the paper's: 0.0053 \u00b1 0.0039 superficially (0.0027 \u00b1 0.0005) and 0.0028 \u00b1 0.0019 deep (0.0023 \u00b1 0.0003). With the rule's b (16 mice) the values are 70.6 Hz, 0.0051 and 0.0026; with the 0.5-3 Hz phase (reading b), 70.7 Hz, 0.0078 and 0.0048. The superficial coupling at about 70 Hz holds; the index values do not. The deep part of the statement does not hold as stated either: below c the index peaks above 100 Hz in only 5 of 18 mice (6 with the 0.5-3 Hz phase), and its peak over 5-200 Hz has a median of 71 Hz; the criterion's deep index, the largest above 100 Hz, falls at a median of 185 Hz (range 105-200).", "deviations": [ "Tort et al. (2008) modulation index with 18 phase bins; the phase band is band-passed (3rd-order Butterworth, zero phase) and its phase taken by the Hilbert transform; amplitudes are Morlet (w0 = 6) amplitudes at 24 log-spaced frequencies from 5 to 200 Hz on every other site; up to 600 s of the state per mouse; the index is averaged over the sites of each region before the peak is taken. The surrogate test of the STAR Methods is not run, as no criterion uses it.", "The layer 6 phase site is 15 sites (300 um) below c, the MUA peak of mid layer 5, an assumption: landmark d cannot be placed by a rule the paper gives, and the coherence boundary that opens the sixth layer lies a median 12.5 sites below c (fig2b). In YMV09, where c is at site 56, the site is clipped to the deepest site (7 below c).", "Sites above landmark b are read as the sites above the layer 4 site of the ICs; the rule's b is a variant. The layer 4 site is the peak loading site of the third of the six highest-variance gamma ICs ordered by depth, an assumption: landmark b cannot be placed by a rule the paper gives, and Figures S3B and S3D label the six LFP layers L2, L3, L4, L5a, L5b, L6, so the third is layer 4.", "YMV19 has no state table and is left out." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "slow-gamma-superficial-wake", "statement": "Slow gamma (30-60 Hz) dominates in superficial layers, most in waking, then REM, then non-REM sleep.", "source": "Results, Brain State-Dependent Features of LFPs, paragraph 1; Figures 3, S3B, S3D", "quote": "Slow gamma oscillations (30\u201360 Hz) dominated in superficial layers, especially in the waking state followed by REM and non-REM sleep", "statistics": [ { "test": "not named", "p": 0.028 } ], "kind": "finding", "inferred": "", "outcome": "reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "Over coherence clusters 1-2, 30-60 Hz power is highest in waking, then REM, then non-REM (medians 37.8, 30.7 and 24.2 uV^2/Hz, 18 mice); Kruskal-Wallis p = 0.0025 (reading a); signed-rank waking against REM p = 0.00025 (16 of 18 mice) and REM against non-REM p = 0.00053 (17 of 18) (reading b). The criterion tests only this state order. The statement's \"dominates in superficial layers\" does not hold here, as in waking, 30-60 Hz power in coherence cluster 1 lies below the mean over sites (median difference -7.06 uV^2/Hz).", "deviations": [ "Clusters are the waking six-cluster coherence solution (fig2b reading b); power as in fig3c-slow-oscillation-nrem-deep.", "YMV19 has no state table and is left out." ], "assumed": false, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig3c-3-6hz-layer4", "statement": "The most prominent LFP pattern in the waking mouse is a 3-6 Hz oscillation with largest power in layer 4.", "source": "Results, Brain State-Dependent Features of LFPs, paragraph 2; Figure 3C, level b", "quote": "The most prominent LFP pattern in the waking mouse was a 3\u20136 Hz oscillation with largest power in layer 4", "statistics": [ { "test": "Kruskal-Wallis test", "p": 2e-06, "n": "19 mice" } ], "kind": "finding", "inferred": "", "outcome": "reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "Mean-subtracted waking 3-6 Hz power differs across the six layers (Kruskal-Wallis p = 0.022 with coherence layers, 0.0013 with IC layers) and is highest in the third layer under both (medians 289 and 224 uV^2/Hz above the site mean). The layer holding the rule's landmark b ranges from the first to the fifth across mice, so the reading through b is not run.", "deviations": [ "Layer 4, defined in the criterion as the layer containing landmark b, is read as the third of the six layers from the surface, as Figures S3B and S3D label them (L2, L3, L4, L5a, L5b, L6); an assumption, as b cannot be placed by a rule the paper gives.", "Power as in fig3c-slow-oscillation-nrem-deep, over up to 600 s of waking.", "YMV19 has no state table and is left out." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig3d-3-6hz-cross-frequency", "statement": "In waking the 3-6 Hz rhythm is phase-coupled to a 15-25 Hz signal in multiple layers (22 \u00b1 5 Hz, modulation index 4.8 \u00b1 1.2 \u00d7 10^-4), to high gamma in superficial layers (90 \u00b1 13 Hz, 2.7 \u00b1 0.7 \u00d7 10^-4) and to the epsilon band (>100 Hz) in layer 5 (6.4 \u00b1 1.8 \u00d7 10^-4).", "source": "Results, Brain State-Dependent Features of LFPs, paragraph 2; Figure 3D, right", "quote": "The 3\u20136 Hz band was cross-frequency phase-coupled to a 15\u201325 Hz band signal in multiple layers (22 \u00b1 5 Hz, modulation index 4.8 \u00b1 1.2 \u00d7 10 \u22124 )", "statistics": [], "kind": "finding", "inferred": "", "outcome": "partially_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "With the 3-6 Hz phase of the layer 4 site in all waking, over 18 mice: the peak in 10-40 Hz of the index averaged over all sites is at 22.0 \u00b1 5.2 Hz (22 \u00b1 5) with index 1.4 \u00d7 10^-4 (4.8 \u00b1 1.2 \u00d7 10^-4); the peak in 60-100 Hz above the layer 4 site is at 83.5 \u00b1 11 Hz (90 \u00b1 13) with index 2.6 \u00d7 10^-4 (2.7 \u00b1 0.7 \u00d7 10^-4); the largest index above 100 Hz over coherence clusters 4-5 (layer 5) is 3.1 \u00d7 10^-4 (6.4 \u00b1 1.8 \u00d7 10^-4). Three of the five values are within tolerance; the 15-25 Hz and layer 5 indices are lower. The waking-immobility reading cannot be run (the dataset holds no movement signal).", "deviations": [ "Tort et al. (2008) modulation index with 18 phase bins; the phase band is band-passed (3rd-order Butterworth, zero phase) and its phase taken by the Hilbert transform; amplitudes are Morlet (w0 = 6) amplitudes at 24 log-spaced frequencies from 5 to 200 Hz on every other site; up to 600 s of the state per mouse; the index is averaged over the sites of each region before the peak is taken. The surrogate test of the STAR Methods is not run, as no criterion uses it.", "The layer 4 site is the peak loading site of the third of the six highest-variance gamma ICs ordered by depth, an assumption: landmark b cannot be placed by a rule the paper gives, and Figures S3B and S3D label the six LFP layers L2, L3, L4, L5a, L5b, L6, so the third is layer 4.", "The 15-25 Hz coupling \"in multiple layers\" is read as the peak in 10-40 Hz of the index averaged over all sites; layer 5 as coherence clusters 4 and 5 (L5a, L5b) of the waking six-cluster solution (fig2b).", "YMV19 has no state table and is left out." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "figs2c-3-6hz-theta-anticorrelation", "statement": "The 3-6 Hz rhythm vanishes with movement; its power is negatively correlated with hippocampal theta power.", "source": "Results, Brain State-Dependent Features of LFPs, paragraph 3; Figure S2C", "quote": "The antagonism between movement and the 3\u20136 Hz oscillation was quantified by the negative correlation between its power and the power of theta oscillation (p < 0.01, signed rank test; Figure S2 C)", "statistics": [ { "test": "signed rank test", "p": 0.01, "p_relation": "<" } ], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "In waking, 3-6 Hz power at the layer 4 site correlates positively with 6-9 Hz power at the deepest site (median Spearman rho 0.08, negative in 1 of 18 mice with the wavelet time base, 5 of 18 with 10 s windows) and at the same site (0.45 and 0.41, negative in none). The signed-rank tests reject zero in the positive direction at p < 0.01 in three of the four comparisons; with 10 s windows at the deepest site p = 0.01.", "deviations": [ "The layer 4 site is the peak loading site of the third of the six highest-variance gamma ICs ordered by depth, an assumption: landmark b cannot be placed by a rule the paper gives, and Figures S3B and S3D label the six LFP layers L2, L3, L4, L5a, L5b, L6, so the third is layer 4.", "Theta is read at the deepest site, nearest the hippocampus, as the Results call it volume conducted; the same site is a variant. Wavelet amplitudes (4 frequencies per band) are sampled at 10 Hz; Spearman correlations.", "YMV19 has no state table and is left out." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig3f-units-lock-3-6hz-trough", "statement": "Both principal cells and interneurons are strongly phase-locked to the trough of the layer 4 3-6 Hz rhythm.", "source": "Results, Brain State-Dependent Features of LFPs, paragraph 2; Figure 3F", "quote": "Both principal cells and interneurons were strongly phase-locked to the trough of the layer 4-recorded 3\u20136 Hz rhythm", "statistics": [], "kind": "finding", "inferred": "", "outcome": "reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "Of 965 E cells with at least 50 waking spikes, 639 are significantly locked to the 3-6 Hz LFP of the layer 4 site, with circular mean preferred phase 187 degrees (trough 180; resultant 0.73, Rayleigh p = 1.2e-175); of 230 I cells, 193 are locked, mean 182 degrees (resultant 0.80, p = 1.3e-66). Mouse means range from 110 to 236 degrees.", "deviations": [ "The layer 4 site is the peak loading site of the third of the six highest-variance gamma ICs ordered by depth, an assumption: landmark b cannot be placed by a rule the paper gives, and Figures S3B and S3D label the six LFP layers L2, L3, L4, L5a, L5b, L6, so the third is layer 4.", "The phase is that of a Morlet wavelet (w0 = 6) at 4.24 Hz, the geometric centre of 3-6 Hz, with peaks at 0 degrees.", "YMV19 has no state table and is left out." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "figs2d-visual-offset-rebound", "statement": "Visual stimulation blocks an ongoing 3-6 Hz rhythm, and the offset of a stimulus often induces a short rebound train of 3-6 Hz oscillation.", "source": "Results, Brain State-Dependent Features of LFPs, paragraph 3; Figure S2D (cited as S2I)", "quote": "Visual stimulation also blocked an ongoing 3\u20136 Hz rhythm. However, the offset of visual stimulus often induced a short rebound train 3\u20136 Hz oscillation", "statistics": [], "kind": "finding", "inferred": "", "outcome": "blocked", "blocked_by": "data_missing", "notebook": "reproduce_senzai_2019.ipynb", "notes": "Needs the 400 ms flash times across mice; the dataset holds them for 2 of the 13 stimulated mice (YMV05 and YMV08, 498 intervals of 0.4 s each), too few for a test across mice (the smallest two-sided signed-rank p with 2 mice is 0.5). No other file holds visual stimulus times: the groups under acquisition, processing and stimulus of every file are listed in the notebook. The layer 4 site could be taken from the gamma ICs (fig3f-units-lock-3-6hz-trough).", "deviations": [], "assumed": false, "disagreed": false, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig3ab-alpha-analogy", "statement": "The waking 3-6 Hz rhythm has spike and wave waveforms and a depth profile similar to the alpha rhythm in monkey V1.", "source": "Results, Brain State-Dependent Features of LFPs, paragraph 2; Figures 3A, 3B", "quote": "in which the waveforms occasionally split into short \u201cspike\u201d and longer \u201cwave\u201d components with a depth profile distribution similar to the depth distribution of alpha waves (8\u201312 Hz) in monkey V1", "statistics": [], "kind": "finding", "inferred": "", "outcome": "blocked", "blocked_by": "methods_underspecified", "notebook": "reproduce_senzai_2019.ipynb", "notes": "The criterion records this claim as methods_underspecified; the paper compares with published monkey figures (Bollimunta et al. 2008) and gives no measure of similarity. Not run.", "deviations": [], "assumed": false, "disagreed": false, "data": "elsewhere", "where": "Bollimunta et al. 2008, J. Neurosci. 28, 9976-9988, doi:10.1523/JNEUROSCI.2699-08.2008 (published figure, no deposit)", "proposed": null }, { "claim": "fig4a-unit-counts", "statement": "Clustering one session from each of the 19 mice yields 1,472 units, with 251 narrow-waveform putative interneurons and 1,075 wide-waveform putative principal cells.", "source": "Results, Assignment of Neurons to Physiologically Defined Layers, paragraph 1; Figure 4A", "quote": "Unit clustering was performed as described previously ( Fujisawa et al., 2008 ) using one session from each of the 19 mice, yielding a total of 1,472 units", "statistics": [], "kind": "finding", "inferred": "", "outcome": "reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "1,477 units in the files (1,476 of quality good; 1,472 \u00b1 147). From the wide-band mean waveform, as the Results describe (\"the unfiltered waveform was quantified by the trough-to-peak latency\", Assignment of Neurons to Physiologically Defined Layers, paragraph 1), 246 narrow I cells (251 \u00b1 25) and 1,016 wide E cells (1,075 \u00b1 108); 105 positive units, no fiber unit, and 110 units with fewer than 10 spikes in the 60 s used. With a 300 Hz high-pass the counts are 320 and 941, both outside tolerance. At thresholds of 0.50 and 0.60 ms the I count is 215 and 323. Each mouse contributes 5 to 20 I cells.", "deviations": [ "The dataset holds no waveforms; mean waveforms are averaged from 60 s of raw signal per session (two 30 s blocks at 1000 and 3000 s, median 127 spikes per unit), where the paper used all spikes, and 110 units with fewer than 10 spikes there are left unclassified.", "The wide-band signal follows the Results' unfiltered waveform; the STAR Methods do not say whether the waveform is filtered, and the 300 Hz high-pass of Kilosort is shown as a variant, which does not match.", "The fiber rule reads the STAR Methods' kurtosis of the second derivative of the normalized waveform, with units beyond the mean \u00b1 5 SD of the negative units as fibers; none is." ], "assumed": false, "disagreed": false, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig4a-ccg-classification", "statement": "Short-latency CCG interactions identify 347 E and 100 I cells, and these agree with the waveform classes (97.4% of CCG-based E cells are waveform E cells; 98.0% of CCG-based I cells are waveform I cells).", "source": "Results, Assignment of Neurons to Physiologically Defined Layers, paragraph 1; Figure 4A", "quote": "Using these criteria, 347 and 100 units were classified as CCG-based E and I cells, respectively", "statistics": [], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "673 CCG-based E cells and 187 I cells (347 \u00b1 35 and 100 \u00b1 10), 122 of them both; 74.9% of CCG-based E cells are waveform E cells and 72.7% of CCG-based I cells are waveform I cells (80.9% and 79.5% of those with a waveform class), against 97.4% and 98.0%. The hollowed predictor gives 682 and 188. The dataset's own synaptic_effect labels count 444 excitatory and 145 inhibitory units, also above the paper's counts, and agree with this detection for 349 and 116 of them.", "deviations": [ "The CCG half width is 50 ms and the Gaussian kernel is cut at 5 SD with reflected edges, standard choices that do not reach the tested bins.", "The paper's code may apply selections that the STAR Methods do not state; none is applied here." ], "assumed": false, "disagreed": false, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig4de-optotagging", "statement": "Fourteen of 15 optogenetically identified PV cells and 40 of 52 VGAT cells fall in the narrow-waveform inhibitory group.", "source": "Results, Assignment of Neurons to Physiologically Defined Layers, paragraph 2; Figures 4D, 4E", "quote": "Fourteen of 15 optogenetically identified PV cells (n = 11) and 40 of 52 VGAT cells were part of the inhibitory group defined by the spike waveform criterion", "statistics": [], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "In the 11 mice labelled PV-Cre::Ai32 in the files, 62 units are tagged with pulses pooled and 48 of them are waveform I cells (15 \u00b1 1.5 and 14 \u00b1 1.4); in the 3 VGAT-Cre::Ai32 mice, 18 and 14 (52 \u00b1 5.2 and 40 \u00b1 4). Each intensity separately gives 61 and 18. No unit is tagged in other genotypes or with onsets shifted by 0.5 s. Pooled over both lines, 80 units are tagged and 62 (78%) are I cells, against the paper's 67 and 54 (81%). The genotype counts in the files (11 PV, 3 VGAT) differ from the paper's (5 and 9), so the mice cannot be split as the paper split them. Which side is wrong about the genotypes cannot be told from the data; the paper gives no per-mouse list, and the per-mouse tagging counts are in the notebook.", "deviations": [ "Pulses are the dataset's `Laser diode` intervals (10 ms; three amplitude levels of about 496 pulses, read as the three intensities); YMV08 and YMV14 have 992 pulses at some levels.", "The mice are split by the genotype written in the files, which disagrees with the paper's counts." ], "assumed": false, "disagreed": false, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig5b-eight-clusters", "statement": "K-means clustering of the spike-gamma IC phase-coupling profiles of all units gives eight clusters that largely correspond to the anatomical cell body layers.", "source": "Results, Assignment of Neurons to Physiologically Defined Layers, paragraph 3; Figures 5B, 5C", "quote": "We therefore used the spike-gamma phase coupling profile of all units against the six ICs to perform K-means clustering ( STAR Methods ). This procedure yielded eight clusters", "statistics": [], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "1,195 units clustered (965 E, 230 I; 1,326 \u00b1 133). The silhouette (cityblock) over 2 to 15 clusters is highest at 3 (full profile), 2 (largest coupling per IC), 6 (t-SNE of the profile) and 3 (t-SNE of the values per IC), never at 8. In the eight-cluster solution of the full profile the clusters are numbered by depth, but the medians of clusters 1 to 5 all lie between 6.5 and 12 sites above c, so the clusters do not separate depth as Figure 5C does; 38-40% of the units of clusters 1-2 lie above the rule's b and 31-39% of clusters 7-8 below its d. The eight-cluster solution is still used for the layer claims, as the criterion says. On the six ICs confined in depth (fig2c reading b), which leave out the spread deepest IC, the silhouette favours 2 clusters for both the full profile and one value per IC (1,195 units).", "deviations": [ "K-means with the cityblock distance is run as k-medians (L1 assignment, median centres), ten restarts, lowest within-cluster sum.", "Units are those with an E or I waveform class and at least 50 spikes in the waking blocks, in the 18 mice with a state table.", "The gamma ICA (STAR Methods, Independent Component Analysis of LFPs) is fitted on up to 600 s of waking per mouse (the longest `Awake` epochs of at least 30 s), with scikit-learn's FastICA on the first ten PCs of the 30-100 Hz LFP in place of runica (EEGLAB infomax, not available in Python); the whole-session reading of fig2c is not run.", "Depth is taken in sites relative to landmark c (the MUA power peak), in place of the paper's normalized depth, which warps depth between all five landmarks (STAR Methods, Depth normalization) and so needs a, b, d and e, which no rule given in the paper places (fig1g-landmark-distances-consistent). Aligning on c is an assumption.", "Landmarks b and d are those of the rule of fig1g (0-100 ms, one site), not the paper's." ], "assumed": false, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig5a-spike-ic-depth-stratification", "statement": "Unit coupling to the gamma ICs is stratified in depth; above landmark b units couple to slow gamma, in deep layers coupling shifts toward fast gamma.", "source": "Results, Assignment of Neurons to Physiologically Defined Layers, paragraph 3; Figure 5A; Figure S5", "quote": "Around and above depth level b (i.e., layer 3/4 border; first three components), unit-IC coupling of both putative E and I neurons had a peak in the slow gamma range. In deep layers, unit-IC coupling shifted toward the fast gamma band", "statistics": [], "kind": "finding", "inferred": "", "outcome": "partially_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "Coupling is stratified in depth: significantly modulated units below c prefer faster gamma than units above b (IC of largest coupling, 82.7 against 64.2 Hz, rank-sum p = 1.9e-7, 487 and 286 units; IC nearest the unit, 88.1 against 60.2 Hz, p = 5.5e-14), and the same holds with the layer 4 site in place of b (p = 4.3e-6 and 2.9e-11). But the superficial medians, 60.2 to 64.2 Hz, are just above the 30-60 Hz the criterion asks for.", "deviations": [ "The gamma ICA (STAR Methods, Independent Component Analysis of LFPs) is fitted on up to 600 s of waking per mouse (the longest `Awake` epochs of at least 30 s), with scikit-learn's FastICA on the first ten PCs of the 30-100 Hz LFP in place of runica (EEGLAB infomax, not available in Python); the whole-session reading of fig2c is not run.", "Depth is taken in sites relative to landmark c (the MUA power peak), in place of the paper's normalized depth, which warps depth between all five landmarks (STAR Methods, Depth normalization) and so needs a, b, d and e, which no rule given in the paper places (fig1g-landmark-distances-consistent). Aligning on c is an assumption.", "Landmark b is the rule's (fig1g); the layer 4 site of the ICs is a variant." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig6a-firing-rate-layers", "statement": "Principal cell firing rates differ across layers; the fastest groups are in layers 5A and 5B and the slowest in the superficial layers.", "source": "Results, Layer-Specific and Brain State-Dependent Features of Unit Firing, paragraph 1; Figure 6A", "quote": "The fastest firing group corresponded to those in layers 5A and B ( Figure 6 , red and magenta groups), while the slowest ones were the superficial layers", "statistics": [ { "test": "Kruskal-Wallis test", "p": 3.7e-83 } ], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "E cell rates differ across clusters (Kruskal-Wallis p = 1.1e-49 in non-REM, 1.8e-63 over the session; 965 E cells), but the two fastest clusters are 7 and 6 (non-REM medians 3.15 and 2.37 Hz) and the two slowest 1 and 8 (0.71 and 1.48 Hz) under both readings, where the claim has 5 and 6 fastest and 1 and 2 slowest.", "deviations": [ "The eight unit clusters are those of fig5b-eight-clusters (k-medians on the full waking spike-IC coupling profile, numbered by median depth relative to c), used though the silhouette does not favour eight, as the criterion of fig5b says; the clusters are an assumption, and the layer names (layer 2/3 clusters 1-2, layer 4 cluster 3, layer 5 clusters 4-6, layer 6 clusters 7-8) are the claims file's.", "YMV19 has no state table and is left out." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig6b-deep-wake-higher", "statement": "Deep-layer principal neurons fire more during waking than during non-REM sleep.", "source": "Results, Layer-Specific and Brain State-Dependent Features of Unit Firing, paragraph 1; Figure 6B", "quote": "Deep-layer neurons were significantly more active during waking compared with non-REM sleep", "statistics": [], "kind": "finding", "inferred": "", "outcome": "reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "Under the signed-rank reading with deep groups as clusters 5-8, each group fires more in waking than in non-REM (p = 1.5e-5, 2.3e-20, 6.8e-37, 8.9e-13; median ratios 1.13, 1.43, 1.54, 1.36). Cluster 4 does not differ (p = 0.16, 29 cells), so the clusters 4-8 reading fails, and by the rank-sum test cluster 5 does not differ (p = 0.11).", "deviations": [ "The eight unit clusters are those of fig5b-eight-clusters (k-medians on the full waking spike-IC coupling profile, numbered by median depth relative to c), used though the silhouette does not favour eight, as the criterion of fig5b says; the clusters are an assumption, and the layer names (layer 2/3 clusters 1-2, layer 4 cluster 3, layer 5 clusters 4-6, layer 6 clusters 7-8) are the claims file's.", "YMV19 has no state table and is left out." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig6b-superficial-no-state-change", "statement": "No such state dependence of firing rates is present in superficial neurons.", "source": "Discussion, State Dependence of Firing Patterns and Interlaminar Interactions, paragraph 1; Figure 6B", "quote": "Deep-layer neurons increased their excitability during waking ( Sakata and Harris, 2012 ), as quantified by their firing rates, whereas no such state dependence was present in superficial neurons", "statistics": [], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "Cluster 1 differs between waking and non-REM (signed-rank p = 0.009, rank-sum p = 0.048), firing less in waking (median ratio 0.87, 93 cells); cluster 2 does not (p = 0.073 and 0.54; ratio 1.08).", "deviations": [ "The eight unit clusters are those of fig5b-eight-clusters (k-medians on the full waking spike-IC coupling profile, numbered by median depth relative to c), used though the silhouette does not favour eight, as the criterion of fig5b says; the clusters are an assumption, and the layer names (layer 2/3 clusters 1-2, layer 4 cluster 3, layer 5 clusters 4-6, layer 6 clusters 7-8) are the claims file's.", "YMV19 has no state table and is left out." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig6c-burst-index", "statement": "Superficial-layer principal neurons are the most bursty, followed by layer 6 and layer 4, while layer 5 neurons have the lowest burst indices.", "source": "Results, Layer-Specific and Brain State-Dependent Features of Unit Firing, paragraph 2; Figure 6C", "quote": "In general, superficial layer neurons exhibited more bursting behavior, followed by layer 6 and layer 4 neurons, while layer 5 neurons had the lowest burst indices", "statistics": [ { "test": "Kruskal-Wallis test", "p": 5.2e-37 } ], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "Burst indices differ across clusters (Kruskal-Wallis p = 1.3e-41 for reading a, 1.1e-48 for b), and layer 2/3 is the most bursty, but the layer order is 2/3 > 4 > 5 > 6 under both readings (reading b medians 2.99, 2.22, 1.68, 0.93), where the claim has 2/3 > 6 > 4 > 5.", "deviations": [ "The eight unit clusters are those of fig5b-eight-clusters (k-medians on the full waking spike-IC coupling profile, numbered by median depth relative to c), used though the silhouette does not favour eight, as the criterion of fig5b says; the clusters are an assumption, and the layer names (layer 2/3 clusters 1-2, layer 4 cluster 3, layer 5 clusters 4-6, layer 6 clusters 7-8) are the claims file's.", "The autocorrelogram has 0.5 ms bins over the whole session; the ratio of mean counts (reading b) does not depend on the bin width.", "YMV19 has no state table and is left out." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig6d-transient-steady", "statement": "Deep layer 5 neurons show the strongest transient responses after the UP state onset (largest transient/steady firing ratio).", "source": "Results, Layer-Specific and Brain State-Dependent Features of Unit Firing, paragraph 2; Figure 6D", "quote": "By this measure, deep layer 5 neurons showed the strongest transient responses", "statistics": [ { "about": "layer difference", "test": "not named (Kruskal-Wallis by the Methods default)", "p": 5.6e-10 } ], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "The transient/steady ratio differs across clusters (Kruskal-Wallis p = 2.4e-6, 2.2e-6 and 1.4e-6 for 5, 10 and 20 ms bins), but cluster 1 has the highest median under every bin (2.24, 1.97, 1.85), not cluster 6 (1.90, 1.80, 1.73).", "deviations": [ "The eight unit clusters are those of fig5b-eight-clusters (k-medians on the full waking spike-IC coupling profile, numbered by median depth relative to c), used though the silhouette does not favour eight, as the criterion of fig5b says; the clusters are an assumption, and the layer names (layer 2/3 clusters 1-2, layer 4 cluster 3, layer 5 clusters 4-6, layer 6 clusters 7-8) are the claims file's.", "DOWN-UP transitions are the dataset's UP onsets that start where a DOWN interval ends, all of them in each session.", "YMV19 has no state table and is left out." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig6e-down-up-latency", "statement": "During the DOWN-UP transition layer 5B neurons fire first, followed by superficial and layer 6 neurons.", "source": "Results, Layer-Specific and Brain State-Dependent Features of Unit Firing, paragraph 2; Figure 6E", "quote": "During the DOWN-UP state transition, layer 5B neurons fired first, followed by superficial and layer 6 neurons", "statistics": [ { "test": "not named (Kruskal-Wallis by the Methods default)", "p": 1.3e-38 } ], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "Latencies differ across clusters (Kruskal-Wallis p = 2.8e-28 for first spikes, 3e-11 for PETH peaks), but cluster 7 fires first under both readings (median 80.8 ms and 45 ms), with cluster 6 second (92.8 and 50 ms).", "deviations": [ "The eight unit clusters are those of fig5b-eight-clusters (k-medians on the full waking spike-IC coupling profile, numbered by median depth relative to c), used though the silhouette does not favour eight, as the criterion of fig5b says; the clusters are an assumption, and the layer names (layer 2/3 clusters 1-2, layer 4 cluster 3, layer 5 clusters 4-6, layer 6 clusters 7-8) are the claims file's.", "First-spike latencies are taken within 300 ms of each transition; the PETH has 5 ms bins.", "YMV19 has no state table and is left out." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig6f-3-6hz-phase-layers", "statement": "Spike phase preference during the waking 3-6 Hz rhythm differs systematically across layers.", "source": "Results, Layer-Specific and Brain State-Dependent Features of Unit Firing, paragraph 2; Figure 6F", "quote": "The distribution of spike firing latency during the 3\u20136 Hz rhythm of waking displayed characteristic laminar differences", "statistics": [ { "test": "not named", "p": 2.6e-10 } ], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "Preferred 3-6 Hz phases of E cells do not differ across clusters: all 965 cells, Kruskal-Wallis on phases p = 0.55, Watson-Williams p = 0.79; the 639 significantly locked cells, p = 0.084 and 0.36. Cluster mean phases lie between 172 and 196 degrees.", "deviations": [ "The eight unit clusters are those of fig5b-eight-clusters (k-medians on the full waking spike-IC coupling profile, numbered by median depth relative to c), used though the silhouette does not favour eight, as the criterion of fig5b says; the clusters are an assumption, and the layer names (layer 2/3 clusters 1-2, layer 4 cluster 3, layer 5 clusters 4-6, layer 6 clusters 7-8) are the claims file's.", "The layer 4 site is the peak loading site of the third of the six highest-variance gamma ICs ordered by depth, an assumption: landmark b cannot be placed by a rule the paper gives, and Figures S3B and S3D label the six LFP layers L2, L3, L4, L5a, L5b, L6, so the third is layer 4.", "The Watson-Williams test uses the kappa correction of Zar (1999).", "YMV19 has no state table and is left out." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig6g-3-6hz-modulation-strength", "statement": "Phase locking to the 3-6 Hz rhythm is strongest in superficial layers and weakest in layer 6.", "source": "Results, Layer-Specific and Brain State-Dependent Features of Unit Firing, paragraph 2; Figure 6G", "quote": "The mean vector length of phase locking to the 3\u20136 Hz rhythm was strongest in the superficial layers and weakest in layer 6", "statistics": [ { "test": "not named (Kruskal-Wallis by the Methods default)", "p": 7.6e-50 } ], "kind": "finding", "inferred": "", "outcome": "reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "The 3-6 Hz mean resultant length of E cells differs across clusters (Kruskal-Wallis p = 1.1e-48, 965 cells), highest in cluster 2 (median 0.14) and lowest in cluster 7 (0.039), as the criterion asks.", "deviations": [ "The eight unit clusters are those of fig5b-eight-clusters (k-medians on the full waking spike-IC coupling profile, numbered by median depth relative to c), used though the silhouette does not favour eight, as the criterion of fig5b says; the clusters are an assumption, and the layer names (layer 2/3 clusters 1-2, layer 4 cluster 3, layer 5 clusters 4-6, layer 6 clusters 7-8) are the claims file's.", "The layer 4 site is the peak loading site of the third of the six highest-variance gamma ICs ordered by depth, an assumption: landmark b cannot be placed by a rule the paper gives, and Figures S3B and S3D label the six LFP layers L2, L3, L4, L5a, L5b, L6, so the third is layer 4.", "YMV19 has no state table and is left out." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig6h-gamma-coupling-wake", "statement": "Spike-gamma (30-100 Hz) coupling is stronger in waking than in non-REM in all layers except layer 5A.", "source": "Results, Layer-Specific and Brain State-Dependent Features of Unit Firing, paragraph 2; Figure 6H", "quote": "Spike-gamma (30\u2013100 Hz) coupling was stronger in the waking animal compared with non-REM in all layers but layer 5A", "statistics": [ { "test": "rank-sum test", "p": 0.01, "p_relation": "<" } ], "kind": "finding", "inferred": "", "outcome": "partially_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "With the IC nearest each unit at its waking preferred frequency (reading a), gamma coupling is stronger in waking than in non-REM in every cluster (rank-sum p from 1.1e-18 to 5.1e-4, 959 E cells), including cluster 5 (p = 2.4e-7), which the claim excepts. With the largest coupling over the ICs (reading b), waking is stronger in clusters 1-4 only; clusters 5, 6 and 8 do not differ and cluster 7 is stronger in non-REM (p = 1.1e-5). The waking advantage holds in all layers under reading (a), but the layer 5A exception does not.", "deviations": [ "The eight unit clusters are those of fig5b-eight-clusters (k-medians on the full waking spike-IC coupling profile, numbered by median depth relative to c), used though the silhouette does not favour eight, as the criterion of fig5b says; the clusters are an assumption, and the layer names (layer 2/3 clusters 1-2, layer 4 cluster 3, layer 5 clusters 4-6, layer 6 clusters 7-8) are the claims file's.", "Non-REM phases are taken on the ICs of the waking ICA, its unmixing applied to up to 600 s of non-REM; E cells need at least 50 spikes in each state.", "The gamma ICA (STAR Methods, Independent Component Analysis of LFPs) is fitted on up to 600 s of waking per mouse (the longest `Awake` epochs of at least 30 s), with scikit-learn's FastICA on the first ten PCs of the 30-100 Hz LFP in place of runica (EEGLAB infomax, not available in Python); the whole-session reading of fig2c is not run.", "YMV19 has no state table and is left out." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "figs7-strongest-3-6hz-coupling", "statement": "The strongest spike-LFP coupling is in the 3-6 Hz band, followed by the slow gamma (30-60 Hz) band.", "source": "Results, Layer-Specific and Brain State-Dependent Features of Unit Firing, paragraph 2; Figures S7A to S7D", "quote": "The strongest spike-LFP coupling was observed in the 3\u20136 Hz band ( Figures 3 F and S7 ), followed by the slow gamma (30\u201360 Hz) band", "statistics": [], "kind": "finding", "inferred": "", "outcome": "partially_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "With the LFP 10 sites (200 um) from each unit, the mean modulation strength is largest in 3-6 Hz (0.073), then 0.5-3 Hz (0.065); 30-60 Hz is fifth (0.049). By preferred frequency, 0.5-3 Hz leads (591 of 1,270 significantly modulated units), then 3-6 Hz (344); 30-60 Hz has 65. With the LFP of the unit's own site, 60-100 Hz leads under both readings (714 units; mean 0.108), as expected when the unit's spike leaks into its own LFP. With the gamma bands on the gamma IC of largest coupling (the Methods' \"LFP (or ICs)\"), 60-100 Hz (0.118) and 30-60 Hz (0.098) lead and 3-6 Hz (0.073) is third. 3-6 Hz coupling is the strongest under one reading, but under no reading is 30-60 Hz second to it; the Figure S7C legend itself ranks 3-6 Hz and below 2 Hz highest. The LFP site is an assumption; the question for the authors is in the findings. Measured values are the mean modulation strength at 3-6 and 30-60 Hz (distant site).", "deviations": [ "The LFP site is not given by the STAR Methods or the Figure S7 legend. The site 10 sites from the unit's peak site, toward the longer side of the probe, is adopted as an assumption, so that the unit's spike is not in the LFP; the unit's own site is shown and does not match either.", "Waking is up to 600 s per session in total from the longest `Awake` epochs of at least 30 s, spikes within 8 s of an epoch edge left out (beyond \u00b14 SD of the 0.5 Hz wavelet), the LFP decimated to 312.5 Hz, 25 log-spaced frequencies from 0.5 to 100 Hz, Morlet w0 = 6 (Torrence and Compo 1998, cited by the STAR Methods); 120 units with fewer than 50 waking spikes and YMV19 (no state table) are left out.", "Significance is the Rayleigh test at each unit's preferred frequency, without correction over the 25 frequencies, as the STAR Methods describe it.", "The IC reading uses the 20 scales from 30 to 100 Hz of the gamma ICA (fig2c) for the two gamma bands and the distant LFP for the bands below 30 Hz." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig7-connection-counts", "statement": "Of 102,332 potential neuron pairs, 525 putative excitatory and 323 putative inhibitory monosynaptic connections were identified.", "source": "Results, Within-Layer and Inter-layer Interactions among Neurons, paragraph 1", "quote": "Of the 102,332 potential neuron pairs in all mice and sessions, we identified putative excitatory (E, n = 525) and inhibitory (I, n = 323) monosynaptic connections", "statistics": [], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "1,295 excitatory and 578 inhibitory connections (525 \u00b1 53 and 323 \u00b1 32); 1,328 and 591 with the hollowed predictor. Potential pairs are 121,842 ordered among all units, 89,046 ordered among E and I units, 60,921 and 44,523 unordered; none is within 102,332 \u00b1 10,233. YMV07 alone gives 235 excitatory and 96 inhibitory connections. 482 excitatory connections join units 0 or 1 site apart; without them 813 and 462 remain, still above the paper's counts. Detected within each state's CCG, there are 291 excitatory and 138 inhibitory connections in waking and 321 and 236 in non-REM (a diagnostic; the claim counts whole-session connections).", "deviations": [], "assumed": false, "disagreed": false, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig7-within-layer-fractions", "statement": "Many connections are within the same layer (E-I 173 of 385, E-E 19 of 90, I-E 98 of 323), but there are also many cross-layer excitatory and inhibitory pairs.", "source": "Results, Within-Layer and Inter-layer Interactions among Neurons, paragraph 1; Figures 7A, 7B, S8A", "quote": "Although many identified excitatory connections were detected within the same layer (layers in these calculations refer to IC-based clusters) for both E-I ( Figures 7 A and 7B; n = 173 of 385 connections) and E-E (n = 19 of 90) pairs", "statistics": [], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "Within-cluster fractions: E-I 210 of 533 (0.394; 0.404 to 0.494), E-E 144 of 351 (0.410; 0.190 to 0.232), I-E 132 of 360 (0.367; 0.273 to 0.333), counting connections whose units both have a cluster. Without units 0-1 sites apart, 0.402, 0.269 and 0.364, all three still outside (0.402 is just below 0.404). The detection finds 549 E-I, 371 E-E and 372 I-E connections against the paper's 385, 90 and 323.", "deviations": [ "The eight unit clusters are those of fig5b-eight-clusters (k-medians on the full waking spike-IC coupling profile, numbered by median depth relative to c), used though the silhouette does not favour eight, as the criterion of fig5b says; the clusters are an assumption, and the layer names (layer 2/3 clusters 1-2, layer 4 cluster 3, layer 5 clusters 4-6, layer 6 clusters 7-8) are the claims file's.", "Connections are those of fig7-connection-counts (STAR Methods rule, unhollowed)." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig7eg-l23-targets", "statement": "Of the 172 putative excitatory connections of layer 2/3 pyramidal cells, 10 (5.8%) and 39 (22.7%) target layer 2/3 and layer 4/5/6 excitatory neurons, and 29 (16.9%) and 94 (54.7%) target layer 2/3 and layer 4/5 interneurons.", "source": "Results, Within-Layer and Inter-layer Interactions among Neurons, paragraph 1; Figures 7E, 7G", "quote": "Of the 172 putative excitatory connections of L2/3 pyramidal cells, 10 (5.8%) and 39 (22.7%) targeted L2/3 and L4/5/6 excitatory neurons, respectively, in addition to contacting 29 (16.9%) and 94 (54.7%) L2/3 and L4/5 interneurons, respectively", "statistics": [], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "Layer 2/3 E cells (clusters 1-2) make 144 excitatory connections (172 \u00b1 17): 13.9% to layer 2/3 E cells (5.8 \u00b1 0.58), 30.6% to layer 4/5/6 E cells (22.7 \u00b1 2.27), 34.7% to layer 2/3 I cells (16.9 \u00b1 1.69), and 13.9% (clusters 3-6) or 18.8% (clusters 3-8) to layer 4/5 I cells (54.7 \u00b1 5.47); 3 go to units without a cluster.", "deviations": [ "The eight unit clusters are those of fig5b-eight-clusters (k-medians on the full waking spike-IC coupling profile, numbered by median depth relative to c), used though the silhouette does not favour eight, as the criterion of fig5b says; the clusters are an assumption, and the layer names (layer 2/3 clusters 1-2, layer 4 cluster 3, layer 5 clusters 4-6, layer 6 clusters 7-8) are the claims file's." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig7c-divergence-convergence", "statement": "Most presynaptic principal neurons have a single target in the recorded population, a few target several; an interneuron typically has one presynaptic partner, a few many, up to 16.", "source": "Results, Within-Layer and Inter-layer Interactions among Neurons, paragraph 2; Figures 7B, 7C", "quote": "A typical interneuron typically had one presynaptic partner, but a small fraction many, up to 16", "statistics": [], "kind": "finding", "inferred": "", "outcome": "partially_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "549 E-I connections (waveform E to waveform I). Targets per presynaptic E cell have mode 1 (242 of 362 cells) and presynaptic E cells per I cell have mode 1 (39 of 138), as the claim says, but the largest convergence is 29 (an I cell in YMV07; the next is 18), outside 16 \u00b1 1.6. The E-I connections rest on the detection of fig7-connection-counts, which finds more connections than the paper.", "deviations": [], "assumed": false, "disagreed": false, "data": "dataset", "where": null, "proposed": null }, { "claim": "figs8b-reciprocal-connections", "statement": "Almost half of E-I connections (168 of 392, 42.9%) have a reciprocal I-E connection, and reciprocal E-I-E connections are almost twice as frequent in layer 5 (60-70%) as in other layers.", "source": "Results, Within-Layer and Inter-layer Interactions among Neurons, paragraph 2; Figure S8B (the text cites S7C)", "quote": "Almost half of the E-I connections (168 of 392 [42.9%]) received reciprocal I-E connections when all layers were considered. Reciprocal E-I-E connections were almost twice more frequent in layer 5 (60%\u201370%) than in other layers", "statistics": [], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "175 of 549 E-I connections (31.9%) have a reciprocal I-E connection (42.9 \u00b1 4.29%). By cluster of the E cell the fractions are 23.3, 34.0, 34.4, 48.5, 26.7, 20.6, 38.2 and 26.7%; clusters 4-6 pooled give 30.3% (54 to 77%), and the ratio to the other clusters is 0.93 (at least 1.8). 48 of the 175 reciprocal pairs join units 0-1 sites apart.", "deviations": [ "The eight unit clusters are those of fig5b-eight-clusters (k-medians on the full waking spike-IC coupling profile, numbered by median depth relative to c), used though the silhouette does not favour eight, as the criterion of fig5b says; the clusters are an assumption, and the layer names (layer 2/3 clusters 1-2, layer 4 cluster 3, layer 5 clusters 4-6, layer 6 clusters 7-8) are the claims file's." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig7d-l23e-l5i-weaker-nrem", "statement": "Spike transmission from layer 2/3 E cells to layer 5 I cells is weaker during non-REM sleep than during waking.", "source": "Results, Within-Layer and Inter-layer Interactions among Neurons, paragraph 3; Figure 7D", "quote": "We observed that spike transmission strength between layer 2/3 E cells and deep layer I and E cells became weaker during non-REM sleep compared with waking (L2/3 E cells to L5 I cells, p = 0.0012", "statistics": [ { "test": "signed rank test", "p": 0.0012 } ], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "13 E-I connections from clusters 1-2 to I cells in clusters 4-6. Peak-to-predictor ratio, median 1.60 in waking and 1.79 in non-REM, waking larger in 4 (signed-rank p = 0.068); transmission probability 0.0152 and 0.0262, waking larger in 5 (p = 0.31). Neither shows waking > non-REM.", "deviations": [ "The eight unit clusters are those of fig5b-eight-clusters (k-medians on the full waking spike-IC coupling profile, numbered by median depth relative to c), used though the silhouette does not favour eight, as the criterion of fig5b says; the clusters are an assumption, and the layer names (layer 2/3 clusters 1-2, layer 4 cluster 3, layer 5 clusters 4-6, layer 6 clusters 7-8) are the claims file's.", "Each state's CCG uses the presynaptic spikes in that state with its own predictor; the connections are those detected over the whole session.", "YMV19 has no state table and is left out." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig7e-l23e-l4i-weaker-nrem", "statement": "Spike transmission from layer 2/3 E cells to layer 4 I cells is weaker during non-REM sleep than during waking.", "source": "Results, Within-Layer and Inter-layer Interactions among Neurons, paragraph 3; Figure 7E", "quote": "L2/3 E cells to L4 I cells, p = 0.0010", "statistics": [ { "test": "signed rank test", "p": 0.001 } ], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "7 E-I connections from clusters 1-2 to I cells in cluster 3. Ratio medians 1.94 (waking) and 1.86 (non-REM), waking larger in 6 (p = 0.11); transmission 0.029 and 0.027, waking larger in 4 (p = 0.81).", "deviations": [ "The eight unit clusters are those of fig5b-eight-clusters (k-medians on the full waking spike-IC coupling profile, numbered by median depth relative to c), used though the silhouette does not favour eight, as the criterion of fig5b says; the clusters are an assumption, and the layer names (layer 2/3 clusters 1-2, layer 4 cluster 3, layer 5 clusters 4-6, layer 6 clusters 7-8) are the claims file's.", "As fig7d-l23e-l5i-weaker-nrem.", "YMV19 has no state table and is left out." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig7h-l23e-l5e-weaker-nrem", "statement": "Spike transmission from layer 2/3 E cells to layer 5 E cells is weaker during non-REM sleep than during waking.", "source": "Results, Within-Layer and Inter-layer Interactions among Neurons, paragraph 3; Figure 7H", "quote": "L2/3 E cells to L5 E cells, p = 0.015; signed rank test", "statistics": [ { "test": "signed rank test", "p": 0.015 } ], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "28 E-E connections from clusters 1-2 to E cells in clusters 4-6. Ratio medians 1.68 and 1.59, waking larger in 17 (p = 0.28); transmission 0.0158 and 0.0129, waking larger in 15 (p = 0.35).", "deviations": [ "The eight unit clusters are those of fig5b-eight-clusters (k-medians on the full waking spike-IC coupling profile, numbered by median depth relative to c), used though the silhouette does not favour eight, as the criterion of fig5b says; the clusters are an assumption, and the layer names (layer 2/3 clusters 1-2, layer 4 cluster 3, layer 5 clusters 4-6, layer 6 clusters 7-8) are the claims file's.", "As fig7d-l23e-l5i-weaker-nrem.", "YMV19 has no state table and is left out." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig7g-deep-ee-stronger-nrem", "statement": "E-E spike transmission among deep-layer principal cells (layer 5/6 to layer 5/6) is stronger during non-REM sleep than during waking.", "source": "Results, Within-Layer and Inter-layer Interactions among Neurons, paragraph 3; Figures 7G, 7I", "quote": "In contrast, putative E-E spike transmission in the deep layers (layer 5/6 to layer 5/6 principal cells) significantly increased during non-REM sleep compared with waking (p = 0.0061, signed rank test", "statistics": [ { "test": "signed rank test", "p": 0.0061 } ], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "251 E-E connections with both cells in clusters 4-8. Ratio medians 1.478 (waking) and 1.460 (non-REM), waking larger in 116 (p = 0.46); transmission 0.0050 and 0.0054, waking larger in 124 (p = 0.16). Neither shows non-REM > waking.", "deviations": [ "The eight unit clusters are those of fig5b-eight-clusters (k-medians on the full waking spike-IC coupling profile, numbered by median depth relative to c), used though the silhouette does not favour eight, as the criterion of fig5b says; the clusters are an assumption, and the layer names (layer 2/3 clusters 1-2, layer 4 cluster 3, layer 5 clusters 4-6, layer 6 clusters 7-8) are the claims file's.", "As fig7d-l23e-l5i-weaker-nrem.", "YMV19 has no state table and is left out." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "figs8f-ie-dip-wake", "statement": "Over all I-E pairs, spike suppression is stronger in waking than in non-REM sleep (dip 24.1% against 22.8% of the CCG baseline).", "source": "Results, Within-Layer and Inter-layer Interactions among Neurons, paragraph 3; Figure S8F", "quote": "When all I-E pairs were considered, spike suppression probability during waking was significantly stronger compared with non-REM (I-E dip wake, 24.1%; non-REM sleep, 22.8% relative to the CCG baseline", "statistics": [ { "test": "signed rank test", "p": 0.006 } ], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "Over 563 inhibitory connections with spikes in both states (the main reading includes the inhibitory connections onto I cells), mean dips 39.6% in waking and 39.9% in non-REM (medians 35.6% and 36.0%), outside 24.1 \u00b1 2.41% and 22.8 \u00b1 2.28%; waking is not stronger (signed-rank p = 0.37, 265 of 563 pairs larger in waking, median difference -0.14 points). The waveform I to E reading gives 40.0% and 39.4% (p = 0.57). YMV19 has no state table and is left out.", "deviations": [ "I-E connections are read as all inhibitory connections, as the Results count 323 I-E connections (Within-Layer and Inter-layer Interactions among Neurons, paragraph 1), the total of inhibitory connections; the waveform I to E subset is a variant.", "Waking is the dataset's `Awake` intervals; micro-arousals (`MA`) are left out of both states.", "A state's CCG uses the presynaptic spikes in that state, with its own predictor." ], "assumed": false, "disagreed": false, "data": "dataset", "where": null, "proposed": null }, { "claim": "figs8c-lognormal-strength", "statement": "Connection strengths are skewed and lognormally distributed, a few strong connections mixed with many weak ones.", "source": "Discussion, State Dependence of Firing Patterns and Interlaminar Interactions, paragraph 2; Figure S8C", "quote": "The magnitude of both spike transmission probability and single neuron-induced excitatory postsynaptic potentials (EPSPs) (in E-I and E-E connections) and IPSPs (I-E connections) showed a lognormal distribution", "statistics": [], "kind": "finding", "inferred": "", "outcome": "blocked", "blocked_by": "methods_underspecified", "notebook": "reproduce_senzai_2019.ipynb", "notes": "The criterion records this claim as methods_underspecified, as the paper reports no test or measure of lognormality. As description, excitatory transmission probabilities (1,295 connections, median 0.0085) have skewness 2.89, and 0.25 below zero after log10; inhibitory ones (578, absolute values, median 0.0030) 8.45 and 0.05.", "deviations": [], "assumed": false, "disagreed": false, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig8-down-active-neurons", "statement": "Five neurons in four mice are specifically active during the DOWN state of non-REM slow oscillation, all in deep layer 6; four have bursty autocorrelograms and one is regular spiking.", "source": "Results, DOWN State-Active Neurons, paragraph 1; Figures 8A, 8B", "quote": "We found five neurons in four mice that were specifically and dominantly active during the DOWN state of non-REM slow oscillation ( Figures 8 A and 8B). All DOWN state-active neurons were found in deep layers 6", "statistics": [], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "Reading (a), DOWN rate above UP rate, finds 25 units in 13 mice; reading (b), a PETH that peaks before the DOWN-UP transition and falls after it, 22 units in 11 mice; against 5 \u00b1 1 neurons in 4 \u00b1 1 mice. Of them 13 and 9 lie in clusters 7-8, 13 and 10 below the rule's d, and 12 in each reading have no cluster (positive or unclassified waveforms, or too few waking spikes). Most fire only slightly faster in DOWN states (DOWN/UP ratios 1.0 to 4.4). The dataset labels 10 units \"down state\", in 6 mice; 5 pass reading (a). A stricter rule shown as a variant, which is not a reading of the criterion, finds 5 units in 4 mice with a DOWN rate more than twice the UP rate.", "deviations": [ "Reading (b) is made precise as a PETH (5 ms bins, -300 to +300 ms) with its maximum before zero and a higher mean in -100 to 0 ms than in 0 to 100 ms.", "All units are considered, whatever their waveform class.", "The eight unit clusters are those of fig5b-eight-clusters (k-medians on the full waking spike-IC coupling profile, numbered by median depth relative to c), used though the silhouette does not favour eight, as the criterion of fig5b says; the clusters are an assumption, and the layer names (layer 2/3 clusters 1-2, layer 4 cluster 3, layer 5 clusters 4-6, layer 6 clusters 7-8) are the claims file's." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig8-down-up-rates", "statement": "The DOWN state-active neurons fire at 35.2 \u00b1 23.1 Hz during DOWN states and 4.3 \u00b1 3.1 Hz during UP states.", "source": "Results, DOWN State-Active Neurons, paragraph 1; Figure 8B", "quote": "These five neurons were not only active during the DOWN state but mainly silent during the UP state (35.2 \u00b1 23.1 Hz during DOWN, 4.3 \u00b1 3.1 Hz during UP; p < 0.016, rank sum test)", "statistics": [ { "test": "rank sum test", "p": 0.016, "p_relation": "<", "n": "5 neurons" } ], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "The 25 cells of reading (a) fire at 8.9 \u00b1 11.1 Hz in DOWN and 5.2 \u00b1 5.5 Hz in UP states (35.2 \u00b1 23.1 and 4.3 \u00b1 3.1); the UP rate is within tolerance, the DOWN rate is not, and the rank-sum test does not separate them (p = 0.42). Reading (b) gives 10.0 and 6.0 Hz (p = 0.44). The 5 cells of the stricter variant fire at 25.5 \u00b1 10.8 and 8.3 \u00b1 1.2 Hz.", "deviations": [ "Rates are over the dataset's DOWN and UP intervals." ], "assumed": false, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig8cd-down-active-suppressed", "statement": "DOWN state-active neurons are suppressed during light flashes, when other neurons respond, and during optogenetic activation of PV or VGAT neurons, after which they rebound about 30 ms later while other principal cells are still suppressed.", "source": "Results, DOWN State-Active Neurons, paragraph 1; Figures 8C, 8D", "quote": "During light flash presentation, their spiking activity was further reduced at the time when the remaining neurons robustly responded", "statistics": [], "kind": "finding", "inferred": "", "outcome": "partially_reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "In the PV and VGAT mice (by the file genotypes), 17 of the 20 DOWN state-active cells of reading (a) fire less during the 10 ms pulse than in the -100 to 0 ms baseline, so the per-cell condition fails for 3; on average they rebound at 30-50 ms (6.97 against 5.66 Hz) while the other 685 E cells are still suppressed (0.60 against 3.15 Hz). The mean rebound comes mostly from three cells the dataset labels \"down state\", YMV05 units 40 and 71 (18.8 and 31.6 Hz at 30-50 ms against baselines of 6.2 and 14.0 Hz) and YMV08 unit 60 (14.3 against 7.2 Hz); no other cell exceeds its baseline by more than 2 Hz at 30-50 ms. The three such cells recorded with flashes (YMV05, YMV08) fire less in the 100 ms after onset than before (6.9 against 13.4, 5.7 against 10.9, 1.2 against 2.7 Hz); the other 11 stimulated mice have no flash times. The cell set itself (fig8-down-active-neurons) is far larger than the paper's.", "deviations": [ "Mice are split by the genotype in the files (fig4de-optotagging); pulses are the `Laser diode` intervals pooled over intensities.", "Flash onsets are the 0.4 s `Visual laser` intervals of YMV05 and YMV08." ], "assumed": false, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig8eh-down-active-inverse-correlation", "statement": "DOWN state-active neurons are inversely correlated with other principal cells and interneurons in both non-REM sleep and waking.", "source": "Results, DOWN State-Active Neurons, paragraph 1; Figures 8E to 8H", "quote": "Cross-correlation between the DOWN state-active cells and other principal cells showed an inverse correlation not only during non-REM but also in the waking animal", "statistics": [], "kind": "finding", "inferred": "", "outcome": "reproduced", "blocked_by": null, "notebook": "reproduce_senzai_2019.ipynb", "notes": "Over 1,366 D-E pairs, the mean CCG within \u00b150 ms is below its level at 0.5-1 s in both states (mean ratio 0.945 in waking, 0.867 in non-REM; D-I pairs 0.952 and 0.914). The five cells of reading (a) that the dataset labels \"down state\" have the five lowest waking ratios (0.49 to 0.86) and five of the six lowest non-REM ratios (0.42 to 0.77); the unlabelled YMV10 unit 36 (0.74 in non-REM) lies below the labelled YMV15 unit 33 (0.77). Many cells of reading (a) give ratios near 1.", "deviations": [ "CCGs have 10 ms bins over \u00b11 s, with the reference spikes of the DOWN state-active cell in `Awake` or `Non-REM` intervals; cells are reading (a) of fig8-down-active-neurons." ], "assumed": false, "disagreed": true, "data": "dataset", "where": null, "proposed": null } ], "findings": [ { "kind": "data_missing", "statement": "The 520 nm flash times behind the visual evoked CSD (Figure 1E, 1F, S2D) are in 2 of the 13 stimulated mice, as the `Visual laser` tables of YMV05 and YMV08 (498 intervals of 0.4 s each). The other files hold no visual stimulus times.", "found": null, "status": null }, { "kind": "metadata_error", "statement": "The `Visual laser` tables (\"Laser pulses for subject stimulation\") of YMV05 and YMV08 also hold the optogenetic pulses of the `Laser diode` table (1,488 and 2,715 of their intervals start within 5 ms of a `Laser diode` pulse), and one YMV05 interval lasts 49.6 s.", "found": null, "status": null }, { "kind": "metadata_error", "statement": "The genotypes in the files are 11 PV-Cre::Ai32 and 3 VGAT-Cre::Ai32 mice; the paper's STAR Methods give 5 PV-Cre::Ai32 and 9 VGAT-Cre::Ai32. The other genotypes (3 Ai35, 1 CaMKII-Cre::Ai35, 1 CaMKII-Cre::Ai32) agree.", "found": null, "status": null }, { "kind": "metadata_insufficient", "statement": "The electrode table gives no depth or position (x, y and z are NaN, location \"unknown\", impedance -1); only rel_y (-1 to -64) implies the site order. The data are consistent with the columns being in that depth order (for each file the best-correlated site in the 100-500 Hz LFP is an adjacent one for 50 to 63 of the 64 sites). The units carry no waveform, layer, depth, landmark or cluster label, so every laminar analysis of the paper must be rebuilt from the raw signal and LFP.", "found": null, "status": null }, { "kind": "metadata_insufficient", "statement": "The units' `max_channel` holds the electrode table `id` (a hardware channel number; the ids are a permutation of 0 to 63) of the unit's peak site: it equals the id of the column of the unit's largest mean waveform for 1,157 of 1,255 units (1,247 within one site), and the column index for 76. The column description (\"Channel that observed the maximum waveform amplitude\") does not say which numbering it uses.", "found": null, "status": null }, { "kind": "metadata_insufficient", "statement": "The `cell_type` labels (Pyramidal Cell 836, Narrow Interneuron 277, Wide Interneuron 223, Unknown 15, down state 10) follow a scheme the paper does not describe; 198 \"Wide Interneuron\" units have a waveform E trough-to-peak latency. The 10 \"down state\" units are in 6 mice, where the paper reports 5 DOWN state-active neurons in 4 mice. The `synaptic_effect` labels (444 Excitatory, 145 Inhibitory) do not match the paper's 347 CCG-based E and 100 I cells, and their rule is not given.", "found": null, "status": null }, { "kind": "other", "statement": "YMV03 has no unit columns but quality (no cell_type, synaptic_effect, max_channel or firing_rate), and one of its units is \"unsorted\". YMV19 has no Sleep states table.", "found": null, "status": null }, { "kind": "other", "statement": "YMV07, a CaMKII-Cre::Ai35 mouse, has 41,022 laser pulses of 20 ms at one level, a protocol the paper does not describe.", "found": null, "status": null }, { "kind": "methods_unavailable", "statement": "Question for the paper's authors. How were the CSD landmarks a, b, d and e chosen on the DOWN-UP CSD map? The paper calls them \"the most prominent\" sinks and sources and marks them by hand in Figure 1D. A rule that takes the extreme CSD above and below the MUA peak places them inconsistently (adjacent distances ranging by 220 to 620 um across mice), so normalized depth and every layer group of the paper cannot be rebuilt from the text.", "found": null, "status": null }, { "kind": "methods_unavailable", "statement": "Question for the paper's authors. Which LFP site gives the phases of the spike-LFP coupling of Figure S7 (the unit's own site, a fixed layer, or an IC)? With the unit's own site the unit's spike dominates the high frequencies; 10 sites away 3-6 Hz is strongest but 0.5-3 Hz is second and 30-60 Hz fifth.", "found": null, "status": null }, { "kind": "methods_unavailable", "statement": "Question for the paper's authors. Were further selections applied in the monosynaptic detection (for example on pairs recorded on the same or neighbouring sites, on firing rates, or by refractory contamination)? The rule of the STAR Methods applied to these units finds 1,295 excitatory and 578 inhibitory connections and 673 and 187 CCG-based E and I cells, about twice the paper's 525, 323, 347 and 100, and the pair count of 102,332 matches no count of pairs of the 1,477 units.", "found": null, "status": null }, { "kind": "methods_unavailable", "statement": "Question for the paper's authors. How were the eight unit clusters obtained: on the spike-IC coupling matrix or on its t-SNE embedding, with which features per unit (the full 6 IC x 20 scale profile or one value per IC), and over which range of cluster numbers was the silhouette compared? Under each of these readings the silhouette here favours 2 to 6 clusters, never 8, and the eight-cluster solutions do not separate depth as Figure 5C does.", "found": null, "status": null }, { "kind": "methods_unavailable", "statement": "Question for the paper's authors. Which components count as volume conducted in the 0.34 \u00b1 0.11 of gamma variance, and how were the six main ICs chosen among the ten? Here the ten PCs hold 0.99 of the band variance, the six of highest variance include a deep component that carries up to 0.71 of it, and the spread components carry 0.53.", "found": null, "status": null } ], "reconciliation": null, "verifications": [ { "date": "2026-10-29", "kind": "agent", "by": "Independent agent run by CatalystNeuro (same harness, separate context, did not write the notebook)", "reviewed": "Round 1 (verifier-1): the executed notebook without its conclusions, the helper module and the claims file, as of commit fd30d66e6.", "agreed": 12, "of": 47, "disagreements": [ { "claim": "figs7-strongest-3-6hz-coupling", "outcome": "not_reproduced", "reason": "No reading puts 3\u20136 Hz first and 30\u201360 Hz second. Distant site (a) 0.5\u20133 > 3\u20136 > 60\u2013100, (b) 3\u20136 > 0.5\u20133 > 6\u201315; own site 60\u2013100 first under both readings. Gamma phase is measured on the raw LFP, while the Methods compute it on \"the LFP (or ICs)\" and the paper's gamma coupling is to ICs; the IC reading was not run and could move 30\u201360 Hz up (point B5). COUPLING_EDGE (6 s) is shorter than the wavelet's \u00b14 SD half-length at 0.5 Hz (7.6 s), and the per-epoch 600 s cap lets total waking reach 1142 s (point B8)." }, { "claim": "ica-volume-conduction-fraction", "outcome": "blocked", "reason": "None of these criteria needs landmarks a, b, d or e. Each needs per-mouse ICA or coherence clustering in site coordinates, a coherence-cluster \"layer 4\", or brain states. The \"methods underspecified, inherited\" reason is not in the claims file: `depends_on` records ordering, not untestability, and the size estimate shows size is not the obstacle either (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "fig2d-superficial-ic-power", "outcome": "blocked", "reason": "None of these criteria needs landmarks a, b, d or e. Each needs per-mouse ICA or coherence clustering in site coordinates, a coherence-cluster \"layer 4\", or brain states. The \"methods underspecified, inherited\" reason is not in the claims file: `depends_on` records ordering, not untestability, and the size estimate shows size is not the obstacle either (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "fig2b-six-coherence-clusters", "outcome": "blocked", "reason": "None of these criteria needs landmarks a, b, d or e. Each needs per-mouse ICA or coherence clustering in site coordinates, a coherence-cluster \"layer 4\", or brain states. The \"methods underspecified, inherited\" reason is not in the claims file: `depends_on` records ordering, not untestability, and the size estimate shows size is not the obstacle either (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "fig2b-clusters-same-across-states", "outcome": "blocked", "reason": "None of these criteria needs landmarks a, b, d or e. Each needs per-mouse ICA or coherence clustering in site coordinates, a coherence-cluster \"layer 4\", or brain states. The \"methods underspecified, inherited\" reason is not in the claims file: `depends_on` records ordering, not untestability, and the size estimate shows size is not the obstacle either (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "fig3c-slow-oscillation-nrem-deep", "outcome": "blocked", "reason": "None of these criteria needs landmarks a, b, d or e. Each needs per-mouse ICA or coherence clustering in site coordinates, a coherence-cluster \"layer 4\", or brain states. The \"methods underspecified, inherited\" reason is not in the claims file: `depends_on` records ordering, not untestability, and the size estimate shows size is not the obstacle either (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "slow-gamma-superficial-wake", "outcome": "blocked", "reason": "None of these criteria needs landmarks a, b, d or e. Each needs per-mouse ICA or coherence clustering in site coordinates, a coherence-cluster \"layer 4\", or brain states. The \"methods underspecified, inherited\" reason is not in the claims file: `depends_on` records ordering, not untestability, and the size estimate shows size is not the obstacle either (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "fig3c-3-6hz-layer4", "outcome": "blocked", "reason": "None of these criteria needs landmarks a, b, d or e. Each needs per-mouse ICA or coherence clustering in site coordinates, a coherence-cluster \"layer 4\", or brain states. The \"methods underspecified, inherited\" reason is not in the claims file: `depends_on` records ordering, not untestability, and the size estimate shows size is not the obstacle either (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "fig3d-3-6hz-cross-frequency", "outcome": "blocked", "reason": "None of these criteria needs landmarks a, b, d or e. Each needs per-mouse ICA or coherence clustering in site coordinates, a coherence-cluster \"layer 4\", or brain states. The \"methods underspecified, inherited\" reason is not in the claims file: `depends_on` records ordering, not untestability, and the size estimate shows size is not the obstacle either (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "figs2c-3-6hz-theta-anticorrelation", "outcome": "blocked", "reason": "None of these criteria needs landmarks a, b, d or e. Each needs per-mouse ICA or coherence clustering in site coordinates, a coherence-cluster \"layer 4\", or brain states. The \"methods underspecified, inherited\" reason is not in the claims file: `depends_on` records ordering, not untestability, and the size estimate shows size is not the obstacle either (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "fig3f-units-lock-3-6hz-trough", "outcome": "blocked", "reason": "None of these criteria needs landmarks a, b, d or e. Each needs per-mouse ICA or coherence clustering in site coordinates, a coherence-cluster \"layer 4\", or brain states. The \"methods underspecified, inherited\" reason is not in the claims file: `depends_on` records ordering, not untestability, and the size estimate shows size is not the obstacle either (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "fig2c-six-gamma-ics", "outcome": "blocked", "reason": "These need depth alignment or b. Landmark c was placed in all 19 mice and could serve for alignment; b could be stood in for by coherence cluster 3 (point B1). Its outcome: blocked (other: not run; partly testable)." }, { "claim": "fig3d-slow-oscillation-gamma-coupling", "outcome": "blocked", "reason": "These need depth alignment or b. Landmark c was placed in all 19 mice and could serve for alignment; b could be stood in for by coherence cluster 3 (point B1). Its outcome: blocked (other: not run; partly testable)." }, { "claim": "fig5a-spike-ic-depth-stratification", "outcome": "blocked", "reason": "These need depth alignment or b. Landmark c was placed in all 19 mice and could serve for alignment; b could be stood in for by coherence cluster 3 (point B1). Its outcome: blocked (other: not run; partly testable)." }, { "claim": "fig5b-eight-clusters", "outcome": "blocked", "reason": "The eight unit clusters come from K-means on unit\u2013IC coupling, which needs no CSD landmarks. Only fig5b's depth-order check needs b, c and d. States, UP/DOWN intervals and connections are all loaded (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "fig6a-firing-rate-layers", "outcome": "blocked", "reason": "The eight unit clusters come from K-means on unit\u2013IC coupling, which needs no CSD landmarks. Only fig5b's depth-order check needs b, c and d. States, UP/DOWN intervals and connections are all loaded (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "fig6b-deep-wake-higher", "outcome": "blocked", "reason": "The eight unit clusters come from K-means on unit\u2013IC coupling, which needs no CSD landmarks. Only fig5b's depth-order check needs b, c and d. States, UP/DOWN intervals and connections are all loaded (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "fig6b-superficial-no-state-change", "outcome": "blocked", "reason": "The eight unit clusters come from K-means on unit\u2013IC coupling, which needs no CSD landmarks. Only fig5b's depth-order check needs b, c and d. States, UP/DOWN intervals and connections are all loaded (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "fig6c-burst-index", "outcome": "blocked", "reason": "The eight unit clusters come from K-means on unit\u2013IC coupling, which needs no CSD landmarks. Only fig5b's depth-order check needs b, c and d. States, UP/DOWN intervals and connections are all loaded (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "fig6d-transient-steady", "outcome": "blocked", "reason": "The eight unit clusters come from K-means on unit\u2013IC coupling, which needs no CSD landmarks. Only fig5b's depth-order check needs b, c and d. States, UP/DOWN intervals and connections are all loaded (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "fig6e-down-up-latency", "outcome": "blocked", "reason": "The eight unit clusters come from K-means on unit\u2013IC coupling, which needs no CSD landmarks. Only fig5b's depth-order check needs b, c and d. States, UP/DOWN intervals and connections are all loaded (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "fig6f-3-6hz-phase-layers", "outcome": "blocked", "reason": "The eight unit clusters come from K-means on unit\u2013IC coupling, which needs no CSD landmarks. Only fig5b's depth-order check needs b, c and d. States, UP/DOWN intervals and connections are all loaded (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "fig6g-3-6hz-modulation-strength", "outcome": "blocked", "reason": "The eight unit clusters come from K-means on unit\u2013IC coupling, which needs no CSD landmarks. Only fig5b's depth-order check needs b, c and d. States, UP/DOWN intervals and connections are all loaded (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "fig6h-gamma-coupling-wake", "outcome": "blocked", "reason": "The eight unit clusters come from K-means on unit\u2013IC coupling, which needs no CSD landmarks. Only fig5b's depth-order check needs b, c and d. States, UP/DOWN intervals and connections are all loaded (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "fig7-within-layer-fractions", "outcome": "blocked", "reason": "The eight unit clusters come from K-means on unit\u2013IC coupling, which needs no CSD landmarks. Only fig5b's depth-order check needs b, c and d. States, UP/DOWN intervals and connections are all loaded (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "fig7eg-l23-targets", "outcome": "blocked", "reason": "The eight unit clusters come from K-means on unit\u2013IC coupling, which needs no CSD landmarks. Only fig5b's depth-order check needs b, c and d. States, UP/DOWN intervals and connections are all loaded (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "figs8b-reciprocal-connections", "outcome": "blocked", "reason": "The eight unit clusters come from K-means on unit\u2013IC coupling, which needs no CSD landmarks. Only fig5b's depth-order check needs b, c and d. States, UP/DOWN intervals and connections are all loaded (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "fig7d-l23e-l5i-weaker-nrem", "outcome": "blocked", "reason": "The eight unit clusters come from K-means on unit\u2013IC coupling, which needs no CSD landmarks. Only fig5b's depth-order check needs b, c and d. States, UP/DOWN intervals and connections are all loaded (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "fig7e-l23e-l4i-weaker-nrem", "outcome": "blocked", "reason": "The eight unit clusters come from K-means on unit\u2013IC coupling, which needs no CSD landmarks. Only fig5b's depth-order check needs b, c and d. States, UP/DOWN intervals and connections are all loaded (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "fig7h-l23e-l5e-weaker-nrem", "outcome": "blocked", "reason": "The eight unit clusters come from K-means on unit\u2013IC coupling, which needs no CSD landmarks. Only fig5b's depth-order check needs b, c and d. States, UP/DOWN intervals and connections are all loaded (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "fig7g-deep-ee-stronger-nrem", "outcome": "blocked", "reason": "The eight unit clusters come from K-means on unit\u2013IC coupling, which needs no CSD landmarks. Only fig5b's depth-order check needs b, c and d. States, UP/DOWN intervals and connections are all loaded (point B1). Its outcome: blocked (other: not run; testable)." }, { "claim": "fig8-down-active-neurons", "outcome": "blocked", "reason": "UP/DOWN intervals are loaded, and the dataset even labels 10 \"down state\" units in 6 mice (the paper has 5 in 4). Only the layer part needs clusters. For fig8cd, visual flashes exist only in YMV05 and YMV08 (point B1). Its outcome: blocked (other: not run; testable, fig8cd only in part)." }, { "claim": "fig8-down-up-rates", "outcome": "blocked", "reason": "UP/DOWN intervals are loaded, and the dataset even labels 10 \"down state\" units in 6 mice (the paper has 5 in 4). Only the layer part needs clusters. For fig8cd, visual flashes exist only in YMV05 and YMV08 (point B1). Its outcome: blocked (other: not run; testable, fig8cd only in part)." }, { "claim": "fig8eh-down-active-inverse-correlation", "outcome": "blocked", "reason": "UP/DOWN intervals are loaded, and the dataset even labels 10 \"down state\" units in 6 mice (the paper has 5 in 4). Only the layer part needs clusters. For fig8cd, visual flashes exist only in YMV05 and YMV08 (point B1). Its outcome: blocked (other: not run; testable, fig8cd only in part)." }, { "claim": "fig8cd-down-active-suppressed", "outcome": "blocked", "reason": "UP/DOWN intervals are loaded, and the dataset even labels 10 \"down state\" units in 6 mice (the paper has 5 in 4). Only the layer part needs clusters. For fig8cd, visual flashes exist only in YMV05 and YMV08 (point B1). Its outcome: blocked (other: not run; testable, fig8cd only in part)." } ] }, { "date": "2026-10-30", "kind": "agent", "by": "Independent agent run by CatalystNeuro (same harness, separate context, did not write the notebook)", "reviewed": "Round 2 (verifier-2): the full executed notebook with its conclusions and summary, the README, the helper module, the claims file and this record, as of commit b0066c136.", "agreed": 47, "of": 47, "disagreements": [] } ] } ], "entries": [ { "name": "000166-senzai-2019", "type": "reproduction", "title": "Layer-specific features of mouse V1 units (Senzai et al. 2019)", "description": "Tests the findings of Senzai, Fernandez-Ruiz and Buzsaki (2019) on dandiset 000166, which holds one session from each of 19 mice recorded with a 64-site linear probe in V1, with spike-sorted units, LFP, the raw signal, brain state and UP/DOWN intervals and laser pulse times. It tests the unit counts and waveform classes, the CCG-based detection of monosynaptic connections, the optogenetic tagging, the gamma ICA and coherence clusters, the laminar LFP power, cross-frequency coupling and 3-6 Hz phase locking, the K-means unit clusters and the layer features of firing and connections that rest on them, and the DOWN state-active neurons. Depth is aligned on the MUA power peak, as the CSD landmarks cannot be placed by a rule the paper gives.", "datasets": [ "dandi:000166" ], "papers": [ "10.1016/j.neuron.2018.12.009" ], "derived_from": [], "notebooks": [ { "path": "reproduce_senzai_2019.ipynb", "title": "Reproducing Senzai et al. (2019) from dandiset 000166", "executed": "notebooks/000166-senzai-2019/reproduce_senzai_2019.html" } ] } ] }