DANDI:000336

13 mice in total underwent imaging (L2/3-D: n = 3, L2/3-S: n = 3, L5-D: n = 4, L5-S: n = 3) with at least three optical imaging sessions recorded in each ... the dataset is well-split between male (7) and female (6) subjects

Methods, Experimental animals and calcium imaging, paragraph 3; Table 1; Data Records, Animal and recording session attributes, paragraph 1

Tested on DANDI:000037. 13 mice, 3 L2/3-D, 3 L2/3-S, 4 L5-D and 3 L5-S, 7 male and 6 female, 3 to 6 sessions each. Sex, line (from genotype) and plane (from depth) agree with Table 1 for every subject, and the date of birth implied by each file's age agrees with Table 1.

data for 50 recording sessions total were gathered from 13 animals. Of these, two animals had at least one session that did not meet ... If only the basic data files for sessions 1 to 3 that passed quality control are needed, the total download size is approximately 15 GB for 33 files in total

Data Records, Animal and recording session attributes, paragraph 1; Table 2; Usage Notes, paragraph 1

Tested on DANDI:000037. 50 sessions, the set of Table 2. Under reading (a) 2 mice (411400 and 412933) have a failed session among sessions 1 to 3; under reading (b) 3 mice, adding 440889, whose sessions 5 and 6 failed. 33 sessions 1 to 3 passed QC, from 11 mice.

How the test differed from the paper’s method

• The files carry no QC status or session number. Both come from the paper's code table (mouse_df.csv), which agrees with Table 2 for all 50 sessions on QC, date, depth, ROI count and tracked ROI count (-1 in the table for 0), and whose session numbers run 1, 2, ... in date order within each mouse.

they underwent three 70-minute optical imaging sessions within a span of three to six days, with no more than one session occurring per day

Methods, Experimental animals and calcium imaging, paragraph 1; Fig. 2a

Tested on DANDI:000037. Under readings (a) (sessions 1 to 3) and (ii) (calendar days inclusive), all 13 mice have their three sessions on distinct local dates within 3 to 6 days. Under (i) only 6 of 13 do, as 7 mice were imaged on three consecutive days (2 days from first to third). Under (b) (tracked sessions) 411400 spans 7 calendar days. Durations under (a), reading (x) 70.23 to 70.53 min (median 70.23, given as measured), reading (y) 70.06 to 70.37. The stored start times are UTC with a Pacific offset (see findings); converted to Pacific time, the dates equal Table 2 for all 50 sessions, and the UTC dates for 46.

How the test differed from the paper’s method

• The claim holds under span reading (ii), calendar days inclusive, and not under (i).

In the somatic layers, 15–224 ROIs per mouse per session were identified and retained for analysis, compared to 159–1636 ROIs in the dendritic layers.

Methods, Experimental animals and calcium imaging, paragraph 6; Table 2

Tested on DANDI:000037. Every session's ROI count equals Table 2. Under readings (a) and (b) the somatic counts run from 15 to 224 and the dendritic from 159 to 1636; under (c) the largest somatic count is 227, within the widened range. No ROI of the 33 analysed sessions has NaN or Inf values.

In total, the modified matching algorithm produced ~100–500 highly plausible matched ROIs per plane, i.e., ~32–75% of the theoretical maximum number of trackable ROIs (L2/3-D: n = 254, L2/3-S: n = 261, L5-D: n = 516, L5-S: n = 129)

Methods, ROI tracking across sessions, paragraph 1; Table 2 ("# Tracked ROIs")

Tested on DANDI:000037. Tracked ROIs (non-NaN tracking_id) summed over the 11 analysed mice are 254, 261, 516 and 129; each mouse has the same tracking indices in its three tracked sessions, and every session's tracked count equals Table 2. Fractions of the theoretical maximum run from 32.9 to 75.0% under reading (a), inside 28.8 to 82.5%; from 25.4 to 65.7% under (b) and from 25.2 to 80.5% under (c), outside it.

How the test differed from the paper’s method

• The theoretical maximum is not defined in the paper; the claim holds under reading (a), the sum of the smallest session ROI count of each mouse, and not under (b) or (c).

In this example, four pairwise matches were identified as conflicts and removed, yielding 136 final matches ... enable two of the pairwise matches to be identified as conflicts and removed, yielding 47 final matches

Figure 8a and 8b legends

Tested on DANDI:000037. 413663 has 136 and 411771 47 tracked ROIs in each of their three sessions. The conflicts removed (4 and 2) are not in the files and are not judged.

optical imaging sessions always lasted 70 minutes, comprising 34 minutes of Gabor sequence stimulus and 17 minutes of visual flow stimulus in each direction. Each stimulus period was flanked by 1 or 30 seconds of grayscreen for the habituation and optical imaging sessions, respectively

Methods, Visual stimulation, paragraph 1

Tested on DANDI:000037, the 33 analysed sessions. Gabor period 34.03 to 34.13 min, each flow direction 17.01 to 17.04 min, the four grey periods 29.98 to 30.04 s (medians given as measured), in all 33 sessions. The order of the stimulus periods varies between sessions.

Inconsistent sequences, accounting for approximately 7% of the Gabor sequences and 5% of visual flow stimulus time, only occurred on optical imaging days, and not on habituation days. In particular, each 70-minute imaging session was broken up into approximately 30 blocks, each comprising 30–90 seconds of consistent sequences followed by several seconds of inconsistent sequences

Methods, Visual stimulation, paragraph 4

Tested on DANDI:000037, the 33 analysed sessions. Gabor 6.93% and flow 4.65% under both readings (a) and (b); per session 6.03 to 7.94% and 3.92 to 5.49%. Blocks per session under reading (i), Gabor 31.5 and flow 32.2 on average; reading (ii) gives 63.7 and does not match. From row times, 49 of 2101 consistent runs exceed 90 s by up to 0.22 s and 377 of 1039 Gabor and 347 of 1062 flow inconsistent runs are outside 3 to 6 s and 2 to 4 s by at most 0.06 s; rounded to the sampling grid of the Methods (1.5 s and 1 s), every run is in range. The block count is judged on the mean over sessions; per session it runs from 29 to 34 (Gabor) and 29 to 35 (flow), so 2 (Gabor) and 11 (flow) of 33 sessions fall outside 27 to 33. The run furthest from its grid value, a consistent Gabor run of 88.887 s in session 832883243 (440889), is 0.387 s longer than 88.5 s. Consistent runs exceed their grid values by a median of 0.81 per mille, and all 22 runs more than 0.25 s off lie in the three sessions with the largest median excess (1.24, 1.23 and 0.82 per mille, up to 5.5), which is consistent with the display running slightly slow in those sessions. The outcome rests on two assumptions (see deviations); on row times, or with the block count required in every session, the claim would not hold. Open for the paper's authors - whether the stated ranges refer to the sampled design durations and whether "approximately 30 blocks" describes sessions on average.

How the test differed from the paper’s method

• Assumed that durations are judged on the sampling grid the Methods state ("sampled randomly and uniformly ... across multiples of 1.5 seconds for the Gabor sequence stimulus and of 1 second for the visual flow stimulus", Methods, Visual stimulation, paragraph 4), as the stated ranges describe the design; measured row times run past the grid values by display timing, up to 0.39 s, and on row times 49 consistent, 377 Gabor and 347 flow inconsistent runs fall outside the ranges.
• Assumed that the block count is judged on its mean over sessions, as "approximately 30 blocks" describes sessions in general; per session 2 Gabor and 11 flow counts of 33 fall outside 27 to 33.
• Block counts hold under reading (i), per stimulus, and not under (ii), per session.

Both the somatic and dendritic ROIs showed clear responses to both the Gabor and visual flow stimuli, with many showing increased fluorescence responses to the onset of the stimuli ... There were also clear differences in the responses to the consistent versus inconsistent stimuli as well

Overview of data, paragraph 1; Figure 5a to 5d

Tested on DANDI:000037, session 1 of the 11 analysed mice. Under all four reading pairs each plane has at least one stimulus with a consistent against inconsistent difference. With all ROIs and the Wilcoxon test, L2/3-D differs for Gabor (U minus D -0.0025 dF/F, p = 3.1e-14) and both flow directions; L2/3-S for both flow directions only (Gabor p = 0.72); L5-D for nasal flow only (p = 2.7e-08); L5-S for Gabor (0.017, p = 3.9e-04) and nasal flow. The same holds with the code's scaled traces. Selection (b) picks 0 to 72 ROIs per session and stimulus (none for Gabor in 418779). Selecting Gabor ROIs on single inconsistent sequence responses, as the Fig. 5b legend ("Same as a., but for inconsistent sequences") can be read, gives no Gabor difference in L2/3-D, L2/3-S or L5-D under either test (L5-S, 13 ROIs, Wilcoxon p = 0.011, mean within 2 SEM), but every plane keeps a flow difference, so the outcome is unchanged. The first part of the statement (onset responses) is tested by fig9-onset-offset-transients.

How the test differed from the paper’s method

• Reading (b) for Gabor computes the legend's criteria on single consistent sequence responses from 0.9 s before to 0.6 s after D onset (the Fig. 5a legend's selection); for flow on single inconsistent flow responses from 2 s before to 2 s after onset, with the mean response after minus before onset as the inconsistent minus consistent difference. Under (b) the flow selection uses the tested difference itself, so its test is not independent of the selection.
• The paper's code scales traces for this figure ((trace - median) / (95th - 5th percentile), analysis/session.py); the claims file uses unscaled dF/F, and the scaled result is a variant with the same outcome.

Histograms of aggregated running velocities ( top ) and pupil diameters ( bottom ), for each session. Data is stacked across all mice (11)

Overview of data, paragraph 2; Figure 6b legend; Methods, Running and pupil tracking

Tested on DANDI:000037. All 33 analysed sessions, from 11 mice, have a running velocity trace (cm/s, 60 Hz) and a pupil diameter trace (mm, on the 2p timestamps) spanning the stimulus period; 0.06 to 2.3% of pupil frames are NaN and at most 12 running frames per session are dropped by the 50 cm/s rule.

demonstrate that our recordings have relatively high SNR (>1) and that this is quite stable over days

Technical Validation, paragraph 2; Figure 7a

Tested on DANDI:000037. Group medians 2.37 to 3.43 and 25th percentiles 2.32 to 2.79; no ROI of 10736 has SNR at or below 1. In every plane each session's median lies inside the other sessions' boxes and within 4.8% of the mean of the three medians, so all four reading pairs hold. The threshold of 1 does not separate signal from noise for this measure. Pure Gaussian noise gives an SNR of 2.06 (1.56 and 2.56 with the median 0.5 SD below or above), as the analytic m / sd + 2.063 gives. Every group's median SNR is above its ROIs' own Gaussian floor (median 2.062), by 0.31 to 0.33 in L2/3-D, 0.46 to 0.49 in L5-D and 0.77 to 1.37 in the somatic planes; 3 to 9% of dendritic and 40 to 62% of somatic ROIs exceed it by more than 1. The helper's calculate_snr and rolling mean equal those of the paper's code (util at commit 2b05f954), and its pairwise correlations equal those of the code's own correlation step (get_correlation at commit 046e7fe), on every ROI and pair of 418779 and 420011 session 1 (largest differences 0, under 2e-15 and under 3e-8).

Similarly, the mean ΔF/F signal was stable over days

Technical Validation, paragraph 2; Figure 7b

Tested on DANDI:000037. Both readings hold in all four planes; the largest deviation of a session median from the mean of the three is 9.4% (L5-D), the others 4.0 to 6.7%.

These ΔF/F traces show distinct transients that align with the stimulus onsets and offsets

Technical Validation, paragraph 6; Figure 9a and 9b

Tested on DANDI:000037. Both readings find a transient at onset and at offset in all 12 groups. Onset differences 0.049 to 0.103 dF/F (Wilcoxon p at most 8.1e-29); offset differences -0.013 to -0.035 (p at most 3.2e-25). Each of the 33 mouse sessions has p < 0.05 at both. With Gabor events only, onsets hold in 12 of 12 groups and offsets in 7 (i) and 9 (ii) of 12; with the code's scaled traces, 12 of 12 for both. The Fig. 9 legends name Gabor onsets and offsets only, while the text and the code use both stimuli, as the claims file does; whether the claim should be judged on Gabor events alone is asked of a reader in questions/000336.yaml. On that variant both readings agree that some groups lack an offset transient, which the criterion calls partially reproduced.

How the test differed from the paper’s method

• The paper's code scales traces for this figure; the claims file uses unscaled dF/F, and the scaled result is a variant with the same outcome.

two-photon calcium imaging was performed in either the cell body layer for somatic recordings (175 μ m depth for layer 2/3 and 375 μ m depth for layer 5) or in cortical layer 1 for distal apical dendritic recordings (50–75 μ m depth for layer 2/3 and 20 μ m depth for layer 5) ... Calcium fluorescence movies were recorded at 30 Hz with resonant scanners over a 400 μ m field of view with a resolution of 512 × 512 pixels

Methods, Experimental animals and calcium imaging, paragraphs 2 and 4; Table 2

Tested on DANDI:000037. The imaging rate is 30 Hz in the metadata (30.075 to 30.148 Hz from the timestamps) and every plane image is 512 x 512 pixels. The depth equals Table 2 in 47 of 50 sessions; the three sessions of 433414 (L2/3-D) give 60 um, where Table 2 and the paper's code table give 75 um. The raw imaging plane location string of each of the three files gives depth '60' (area VISp), so 60 um is the deposited value; the code table's depth equals Table 2 in all 50 sessions. The 400 um field of view is not stated in the files. The criterion asks every session's depth to match Table 2, so the claim holds only in part.

The full dataset is publicly available in the Neurodata Without Borders (NWB) format 24 on the DANDI Archive ( https://dandiarchive.org/dandiset/000037 )

Data Records, paragraph 1 and Data organization; Table 2

DANDI:000336 holds 2356 files of 16 subjects and 104 session IDs, none of which is among the 13 subjects and 50 sessions of Table 2; the subject metadata of all 16 subjects (a processed file for 14, a raw file for the 2 without processed files) agree, and their sessions start between 2022-07-25 and 2023-11-15, against 2018-09-26 to 2019-03-18 for Table 2. The archive names 000336 the OpenScope Dendritic Coupling project. All 50 sessions of Table 2 are in DANDI:000037, on which every other claim is tested.

ROIs with correlations above 0.8 (e.g., approximately 0.01% of possible pairs of L2/3 dendrites)

Technical Validation, paragraph 3

Tested on DANDI:000037. Pairs of L2/3-D ROIs with correlation above 0.8 are 0.0178% pooled over the six sessions (reading a), 0.0175% as a mean of sessions (b) and 0.0211% in session 1 (c), all above the 0.015% the tolerance allows. Per session the fraction runs from 0.0018 to 0.046%; 116 of the 134 pairs come from sessions 1 and 2 of 413663. By mouse, 413663 gives 0.032% and 433414 0.0030%; leaving out either mouse gives 0.0030% or 0.032%, both outside the tolerance. The value is of the order the paper gives ("approximately 0.01%") but outside the claims file's tolerance of half its last digit. The helper's correlations equal those of the paper code's get_correlation on every pair of two sessions (fig7a-snr-high-stable).

The distributions of correlation coefficients were very similar for the apical dendrite ROIs and for the somatic ROIs ... suggesting that we were unlikely to be heavily over-segmenting the dendritic data

Technical Validation, paragraph 3; Figure 7c

Tested on DANDI:000037. Reading (ii) holds in all six comparisons (dendritic 99th percentiles 0.081 to 0.119 against somatic 0.175 to 0.207). Reading (i) holds in 5 of 6; in L5 session 3 the dendritic median, 0.0186, lies just below the somatic 25th percentile, 0.0190. The criterion reproduces the claim only when both readings hold in all six comparisons and blocks it when they disagree, so it is blocked with every result shown. Reading (ii) follows the paper's own argument, that over-segmentation would give many highly correlated pairs; on it the claim would be reproduced. The helper's correlations equal those of the paper code's get_correlation on every pair of two sessions (fig7a-snr-high-stable). Open for the paper's authors - what measure of similarity between the dendritic and somatic correlation distributions the statement rests on.

The variation in number of matched ROIs across session orderings for somata was generally far less than that for dendrites due to their larger sizes and more regular shapes.

Technical Validation, paragraph 5; Figure 8a and 8b legends

Tested on DANDI:000037, six orderings each of the three tracked sessions of the 13 tracked mice (sessions 1 to 3 of the 11 analysed mice, sessions 4 to 6 of 411400 and 412933), with results for the 11 analysed mice and for all 13. Somatic mice vary by at most one matched ROI over orderings (range over mean 0 to 0.017, mean 0.004; SD over mean, mean 0.002; 4 of 6 mice do not vary). For the 11 analysed mice, the dendritic mice with the 10% threshold range over mean 0.036 to 0.259 (mean 0.136), SD over mean, mean 0.055; with 20%, 0.053 to 0.405 (mean 0.161) and 0.079. Every dendritic mouse varies more than every somatic one under both thresholds and both measures, and also when the threshold applies only to pairs with a direct pairwise match (range over mean 0.029 to 0.400; means 0.113 at 10% and 0.172 at 20%). Without a threshold the dendritic mean is still larger (0.040 against 0.004), though 433451 (0.015) then varies less than 408021 (0.017). With all 13 tracked mice the dendritic means stay larger under every reading (range over mean 0.104 at 10% and 0.134 at 20%, SD over mean 0.042 and 0.064, against 0.004 and 0.002; 0.034 without a threshold); 412933 matches 60 ROIs in every ordering at 10%, so with it not every dendritic mouse varies more than every somatic one. The somatic match counts (12 to 151) are within 4 of the tracked counts in the files; the dendritic counts under the IoU thresholds (10 to 143 for the analysed mice) are well below them (51 to 353). Per dendritic mouse 19 to 307 three-session matches (on average over orderings) hold a pair of ROIs without a direct pairwise match, against 1 to 4 per somatic mouse; applying the threshold only to linked pairs keeps 3 to 28 more matches per dendritic mouse at 10%, still below the tracked counts (for example 34.3 against 118 for 433414 and 44.2 against 112 for 433448), so the IoU threshold removes most matches. Without a threshold the dendritic counts are 1.3 to 2.6 times the tracked counts. The package's own NwayMatching.run gives the same count as the reconstruction in all 12 orderings of 418779 and 420011. The dendritic values printed in this run differ from those of the run the verifier read in round 2 (for example 433448 range over mean at 10%, 0.259 here and 0.309 there), with the same conclusion; repeating the matching of 420011 and 433414 twice on one machine while preparing (not in the notebook) gave identical counts, so the difference is between runs on different machines. Open for the paper's authors - what overlap measure and threshold the modified matching used for dendrites, and whether the modified code can be published; the Methods say only "overlapped by at least 10-20%", and a measure relative to the smaller ROI would keep more matches.

How the test differed from the paper’s method

• The public package (version 0.6.0, default parameters) in place of the paper's custom-modified version, which is not public, as the claims file directs.
• Assumed that the dendritic overlap is the intersection over union of the registered masks that the package reports, applied to every pairwise match within a three-session match. The paper does not define the overlap; IoU is the overlap measure the package itself computes. The result holds at both thresholds the criterion names.
• The package expects an average intensity projection for registration; the files hold only a max projection, which is used in its place.
• The six directed pairwise matchings of each mouse are computed once and combined per ordering, as NwayMatching.run combines them (itertools.combinations in the given order); the notebook checks this against NwayMatching.run on all six orderings of 418779 and 420011 (12 of 12 equal). The package's summary plots are not made; they fail with the locked pandas (a read-only array in numpy.fill_diagonal) and do not change the matches.
• A pair of ROIs within a three-session match that has no direct pairwise match counts as IoU 0, so the match fails any threshold; the variant that applies the threshold only to linked pairs is shown and gives the same outcome.