DANDI:000233

Notably, SPW-R rates were negatively correlated with future changes in glucose concentrations ... R = -0.21 ± 0.08, n = 30 rats, temporal lag of 10.7 ± 4.5 min

Results, SPW-Rs phase reset glucose oscillations, paragraph 1; Figure 2a-f

Over the 18 rats with CA1 SPW-Rs and glucose recorded together, 3 of the 16 combinations of readings match both expected values. The first the criterion lists (raw delta current, Pearson, one correlogram per rat, lags -100 to +100 min) gives a mean peak R of -0.176 at a mean lag of 14.2 min; the high-pass delta gives -0.180 at 14.4 min, and Spearman with one correlogram per session and the high-pass delta gives -0.184 at 8.8 min. Mean R is within tolerance under all 16 combinations (-0.159 to -0.216) but the mean lag is not under the other 13 (15.8 to 33.1 min). Single-rat lags spread with an s.d. of 26 to 42 min, and leaving out rats with less than 6 h of data leaves 1 matching combination of 16. The peak R is negative in every rat; the median correlogram has its minimum, -0.108, at +5 min. CGM1, the only rat labelled as the ripple and glucose cohort (reading b), peaks at +5 min (R -0.101 Pearson, -0.194 Spearman). The outcome rests on the alignment of the SPW-R count with the glucose change (see deviations); with the count over the 5 min after each reading no per-rat combination matches (mean lags 19.2 to 37.8 min), and with the change to the next reading 4 of 8 match. With all stored events (no 5 s.d. threshold) the Spearman per-rat reading gives R -0.212 at 16.1 min and the Pearson one -0.175 at 14.2 min. The paper's n is 30 rats; 18 is all the dandiset allows. The match is fragile. With at least 6 h of both per rat only the high-pass, Spearman, per-session, -100 to +100 min combination matches. The median of the per-rat lags is 5 to 15 min in every combination, but only 7 to 10 rats peak between +5 and +15 min; in the raw, Pearson, per-rat combination single-rat lags run from -100 to +80 min, and the mean falls in tolerance because peaks at -100 to -20 min offset peaks at +55 to +80 min. R negative in most rats holds by construction, as R is the minimum over 41 offsets.

How the test differed from the paper’s method

• 18 rats where the paper has 30. The dandiset holds 25 subjects; 7 have no SPW-R table or no glucose recorded on the days of the electrophysiology.
• Assumed that the SPW-R count of a 5-min interval is paired with the glucose change over the same interval (the reading at its end minus the one at its start). The paper says only that SPW-R rate and glucose change are correlated at 5-min offsets; this pairing makes offset 0 the same interval for both. The two alignments one reading apart are shown and change the outcome.
• SPW-Rs with a stored peak normed power below 5 s.d. are left out, following the Methods (SPW-R detection, "The peak power threshold was set to five s.d."); the stored tables of nine subjects hold events down to 4 s.d. All stored events are shown as a variant.
• Glucose timestamps of twelve subjects were repaired by swapping day and month in dates whose day is 12 or less (see findings). Without the repair 1139 h of overlap remain of 1717.
• The high-pass reading subtracts a centred 24-h running mean, a standard choice for a cutoff the paper does not give; a session is one file with at least 6 h of bins, a standard choice. Differencing already removes slow drift, so this reading changes delta little and nearly repeats the raw one.
• Reading (a) of the rats leaves out CNO days and the stimulation periods of the optogenetic rats. The files label neither, so every day of every rat is used.

Histograms of the correlation between SPW-R rate and current flux at 10 min (red) and the correlation observed when SPW-R rate is circularly permuted (black) (100 iterations per rat)

Figure 2e-f legends

At +10 min the per-rat Spearman correlations have a median of -0.106 (16 of 18 negative) against 0.005 for 1799 circular shuffles (100 per rat), two-sample KS D = 0.575, p = 4.0e-06. With Pearson, -0.076 against 0.002, p = 1.5e-04. Seven rats (Spearman) lie below all 100 of their own shuffles. The result holds under the two alignments one reading apart, each against its own shuffles. Counting SPW-Rs over the 5 min after each reading, Spearman median -0.108 (KS p = 2.5e-07); taking the change to the next reading, -0.047 (KS p = 2.5e-03). So the outcome does not rest on the alignment assumption.

How the test differed from the paper’s method

• 18 rats where the paper has 30; alignment and 5 s.d. threshold as in fig2-ripple-rate-predicts-glucose.

Isolated SPW-Rs had little or no effect, whereas clustered events with more than 30 SPW-Rs per minute induced a several-fold-larger decrease in peripheral glucose levels

Results, SPW-Rs phase reset glucose oscillations, paragraph 1; Figure 2g

Under reading (a), epochs with more than 150 SPW-Rs are followed 10 min later by a mean delta current of -0.191 nA (585 epochs, 13 rats), against +0.050 nA after epochs with 1 to 5 SPW-Rs (1718 epochs), 3.9 times the magnitude and of opposite sign; a Welch test of the two gives p = 0.059, which the criterion does not require. Under reading (b), more than 30 per epoch, the mean is -0.032 nA, which does not match. The curve is not monotonic (120 to 150 per epoch, +0.125 nA), and in 3 of the 13 rats with clustered epochs their mean is not negative. Under the alignments one reading apart, counting SPW-Rs over the 5 min after each reading gives -0.102 nA after clustered epochs against -0.011 nA after isolated ones under (a), which still matches; taking the change to the next reading gives +0.018 against -0.027 nA under (a) and -0.022 against -0.027 nA under (b), which does not.

How the test differed from the paper’s method

• Epochs pooled over 18 rats; the paper does not say whether it pools.
• Assumed, as in fig2-ripple-rate-predicts-glucose, that the SPW-R count of an epoch is paired with the glucose change over the same interval; the outcome changes under one of the two alignments one reading apart (shown).

Note the slow decay over days, an indication of glucose oxidase enzyme breakdown on the electrode surface ... Note the relatively stable readings over days, and that the slow decay in absolute current readings is no longer present

Extended Data Fig. 1b, 1d

In all 31 placements (runs of at least 3 days, 22 subjects) the daily mean current falls (median -10.7% of the first-day value per day); the daily mean absolute delta current changes by a median -4.7% per day, smaller in magnitude than the current's in 18 of 31 placements.

How the test differed from the paper’s method

• A placement is a run of consecutive readings of at least 3 days, a standard choice; the dandiset does not mark sensor placements.

On average, we observed ... 9.5 (± 8.2) SPW-Rs per minute ... or ~14,000 SPW-Rs per day) with a large degree of variance ranging from zero ripples for ... several minutes up to > 40 ripples in a minute

Supplementary Discussion, paragraph 2; Figure 1f

Under reading (a), 18 rats and 2264 h, SPW-Rs of at least 5 s.d. occur at 10.19 (s.d. 11.15) per minute and 14671 per day, with up to 258 in a minute; minutes with more than 40 occur in 18 of 22 subjects. With all stored events the values are 11.07 per minute and 15941 per day, outside the tolerances. Reading (b), CGM1 alone, gives 10.76 per minute and 15494 per day, outside.

How the test differed from the paper’s method

• SPW-Rs below 5 s.d. left out, as in fig2-ripple-rate-predicts-glucose; the outcome depends on it, and the threshold is the Methods'.

In each animal, we validated that the shape, amplitude, ... and rate of hippocampal SPW-Rs were consistent across multiple days

Supplementary Discussion, paragraph 2; Figure 1d-f

In all 15 rats with at least 3 days of 6 h or more, the daily median amplitude (peak normed power) and duration show no Spearman trend with day at p < 0.05; the daily rate is consistent in 13 of 15 (no trend in 11). Amplitude stays within 20% of the median in only 5 rats and the rate in 4, so the result rests on the absence of trends. The p values are scipy's asymptotic ones, which are too small with few days. A perfectly monotone series of 3 or 4 days gets p = 0.000, where the exact two-sided p is 0.33 (3 days) or 0.083 (4 days), so with 3 or 4 days no series can reach p < 0.05. With exact p no rat with up to 8 days has a trend in any measure; the two rate failures (CGM2, 5 days, exact p 0.083; CGM52, 4 days, exact p 0.083) are artefacts of the asymptotic p, and the rate is consistent in all 15 rats. Amplitude in microvolts trends with day in 5 rats by the asymptotic p and in 2 (CGM1, CGM31, 12 and 18 days) by the exact one; in CGM30 its overall median is 0, which makes its deviation infinite (not used by the criterion).

How the test differed from the paper’s method

• Days with at least 6 h of CA1 recording, a standard choice.

The negative correlation did not change in amplitude across either wake-to-NREM or NREM-to-wake state transitions ... The SPW-R-glucose fluctuation cross-correlogram was not different between NREM and waking states

Extended Data Fig. 4g-i

Three rats (CGM1, CGM2, CGM3) have sleep states with SPW-Rs and glucose. In 4-h blocks the Spearman R at +10 min is -0.017 in NREM (120 blocks) and -0.083 awake (112), unpaired t p = 0.057; per-rat means paired p = 0.27. Across awake-to-NREM transitions (13, two rats) the most negative R from +5 to +20 min is -0.261 before and -0.151 after (p = 0.15); across NREM-to-awake (23, one rat) -0.234 and -0.165 (p = 0.23). No test reaches p < 0.05, with little power at 3 rats. Under the alignments one reading apart, counting SPW-Rs over the 5 min after each reading makes the awake blocks differ from the NREM blocks (-0.109 against -0.043, unpaired p = 0.038), so the claim fails there; taking the change to the next reading, no test reaches p < 0.05 (blocks p = 0.21, transitions p = 0.27 and 0.66).

How the test differed from the paper’s method

• 3 rats where the paper has 8; sleep states are stored for six subjects only.
• Transitions between NREM and awake episodes of at least 10 min each, a choice.
• Spearman correlation, under which the main claim's per-rat reading matched without the alignment variants; the claims file does not fix the type for this claim.
• Assumed, as in fig2-ripple-rate-predicts-glucose, that the SPW-R count of a 5-min interval is paired with the glucose change over the same interval; the outcome changes under one of the two alignments one reading apart (shown).

The most impactful of these was the inter-SPW-R interval (that is, the degree of clustering of SPW-Rs) ... The second most influential feature was the amplitude of SPW-Rs, followed by SPW-R duration; the frequency of the ripple oscillation had the least effect

Results, SPW-Rs phase reset glucose oscillations, paragraph 1; Extended Data Fig. 4a-c

With deciles the ranges of the +10-min effect across bins are interval 0.070 nA, duration 0.069, frequency 0.053 and amplitude 0.040, so the interval leads but the rest of the order differs (the criterion's partial case). With 5-percentile bins duration (0.0967) and interval (0.0966) are tied ahead of amplitude (0.061) and frequency (0.060). Per rat with deciles, amplitude has the largest range in 7 of 18 rats and the interval in 6. 1083233 SPW-Rs from 18 rats. Resampling rats (1000 bootstraps, decile edges fixed), the interval has the largest range in 64.7% of resamples and the full order interval > amplitude > duration > frequency occurs in 1.5%. The most clustered decile (shortest intervals) has the least negative effect, +0.004 nA against -0.027 nA in the decile with the longest intervals, the opposite direction to fig2g-clustered-ripples. Under the alignments one reading apart, taking the change to the next reading keeps the interval first with deciles (0.056 nA), so the outcome is the same; counting SPW-Rs over the 5 min after each reading puts the interval last under both bin widths (deciles frequency 0.060, duration 0.047, amplitude 0.047, interval 0.031), which matches neither case.

How the test differed from the paper’s method

• Assumed, as in fig2-ripple-rate-predicts-glucose, that the 5-min interval holding a SPW-R is paired with the glucose change over the same interval; the outcome changes under one of the two alignments one reading apart (shown).
• The inter-SPW-R interval is the time from the previous SPW-R in the same file, a standard choice (previous or next is not said).

We observed significant power-power and phase-power relationships in the 15-75-min range ... which suggests that a faster SPW-R frequency (about 40 min) modulates a slower (about 60 min) glucose rhythm

Results, SPW-Rs phase reset glucose oscillations, paragraph 2; Extended Data Fig. 5d-e

What the paper reports

• power-power comodulogram against circularly shuffled data: Student's t-test, p < 10⁻²⁴ (n = 30 rats)
• phase of glucose against SPW-R rate amplitude: Rayleigh test for non-uniformity, p < 10⁻¹⁸

Over stretches of at least 12 h with both signals (16 rats), power-power coupling over 15 to 60 min exceeds 100 circular shuffles per rat (mean R 0.064 against -0.007, paired t = 2.57, p = 0.021), and the rate-weighted glucose phases are non-uniform (Rayleigh p = 3.2e-09, resultant length 0.033, above all 200 circular shifts of SPW-R rate, whose 95th percentile is 0.013; p < 0.05 in 10 of 16 rats singly). The comodulogram peak (R 0.216) is on the diagonal at 79.0 min for both signals, so the third condition, a peak at a longer glucose period, fails; the mean correlation where the glucose period is longer than the SPW-R period is 0.072 against 0.047 where it is shorter. With 6-h stretches the same holds (p = 0.025 and 1.1e-08, peak at 79.0 min for both). With 24-h stretches only 3 rats remain and neither test reaches p < 0.05 (p = 0.24 and 0.82). The paper's p bounds (1e-24, 1e-18) are not reached.

How the test differed from the paper’s method

• Periods 12 to 120 min in place of 10 to 120, since 10 min is the Nyquist period of 5-min readings; band edges at the period divided and multiplied by 1.25 (a standard choice).
• Both signals are filtered only over stretches where both are recorded, of at least 12 h (six cycles of the longest period), a choice; 24 h and 6 h are shown, and the 24-h variant, with 3 rats, does not match.

Average power spectral density plot ( n = 8 rats) of ripple rate. Note two peaks at about 24 h and about 4 h

Extended Data Fig. 3a-d; Results, SPW-Rs phase reset glucose oscillations, paragraph 1

The mean Lomb-Scargle spectrum of SPW-R rate of the 8 rats with the most recording has local peaks at 24.0 h and 3.8 h, inside the tolerances, but larger peaks at 42.3 and 19.5 h. Those 8 rats hold 1889.8 h, outside 1568.5 +- 156.85 h. With all 11 rats of at least 48 h (2072.2 h) the nearest peaks are at 24.0 and 4.1 h. By prominence the 24.0-h peak ranks 4th of 45 local peaks and the 3.8-h peak 12th (prominence 0.0034 against 0.0175 for the 42.3-h peak); on a 400-point grid a nearby local peak is weak evidence of a 4-h rhythm.

How the test differed from the paper’s method

• Lomb-Scargle periodogram, a standard choice for gapped recordings; the 8 rats with the most recording, since the dandiset does not say which 8 the paper used.

Furthermore, the phase dispersion of the glucose rhythm decreased after SPW-Rs, and the magnitude of this decrease was proportional to SPW-R rate

Results, SPW-Rs phase reset glucose oscillations, paragraph 2; Figure 2h

5839 peaks of SPW-R rate from 18 rats in deciles of peak rate. The Spearman correlation of the decrease of phase dispersion with the deciles' median rate is rho -0.24 (p = 0.51) with baseline (a) and 0.55 (p = 0.098) with baseline (b), the same for the 15 to 75 and 15 to 90 min bands. The decrease in the top decile is positive under all four readings (0.022 to 0.506), but it is positive in every decile, the lowest included (0.23 to 0.47), so it does not single out high rates. The minimum over seven offsets favours a decrease; with the mean over 0 to +30 min in its place rho is -0.19 to 0.56 (p = 0.09 to 0.86).

How the test differed from the paper’s method

• Band-pass filtering within runs of at least 24 h of readings (second-order Butterworth, zero phase), a standard choice.

Because interstitial glucose fluctuations displayed an approximately 60-min oscillation

Results, SPW-Rs phase reset glucose oscillations, paragraph 2; Extended Data Fig. 1g-i

In 25 subjects the delta-current power above a log-log linear fit is elevated from about 22 to 131 min, with the peak of the mean residual at 51.4 min (per subject 32 to 60 min), outside 60 +- 6 min. Below 22 min the residual is negative, so it is positive at 51 of the 81 frequencies between 15 and 90 min. The first side peak of the mean autocorrelogram is at 70 min (0.013), also outside the tolerance. Neither condition holds under any of 12 variants of the fit's axes (log-log, semi-log, linear), its range (all frequencies or 10 to 300 min) and the segment length (24 or 12 h); the residual peak lies at 43.6 to 51.4 min (log-log), 10.1 min (semi-log) or 68.6 to 72.0 min (linear). The spectrum has a broad hump of excess power from about 22 to 131 min, but the autocorrelogram shows no rhythm at its peak (-0.06 at a 50-min lag, first side peak 0.013 at 70 min); the expected period and the 15 to 90 min range are not matched.

How the test differed from the paper’s method

• All 25 subjects with a 24-h run of readings, where the legend has 8 rats; Welch spectra with 24-h segments, a standard choice (12-h segments shown).

Note the peaks in the histogram from bursts of ripples, and the drop to zero, representing an approximately 20-ms refractory period between ripples

Extended Data Fig. 2a

The shortest interval between consecutive SPW-R peaks within a file, the criterion's reading, is 28.8 to 39.2 ms (median 34.4 ms over 22 rats); no interval below 20 ms occurs and no rat has its shortest interval in 18 to 22 ms. Start-to-start intervals do not match either (median shortest 40.8 ms). The gap from one event's stop to the next one's start has its minimum at exactly 20.0 ms in 19 rats and at 15.2 ms in CGM1, CGM2 and CGM3, so it lies in 18 to 22 ms in 19 of 22 rats. A floor at the same value in nearly every rat matches the merging of neighbouring events in the detection (Methods, SPW-R detection, "merging neighbouring events"), so the 20-ms drop of the legend may be this rule measured from stop to start. The detection parameters are not stored with the NWB tables, so the rule cannot be confirmed.

How the test differed from the paper’s method

• The time point of each event is the claims file's choice (the peak); start-to-start, stop-to-stop and stop-to-next-start are shown as variants. The legend shows auto-correlograms of all inter-ripple intervals; their shortest interval is the same as that of consecutive intervals.

Across rats, no hour of the day had a significantly different correlation than any other hour of the day (Student's t -test; minimum P = 0.53)

Extended Data Fig. 4j-l

What the paper reports

• smallest p over all pairs of hours: Student's t-test, p = 0.53

Over 12 rats with at least 24 h of both signals, Student's t-test finds 10 of 276 pairs of hours at p < 0.05 (smallest p = 0.010), where the paper reports a smallest P of 0.53; the paired variant finds 25. 3 of 288 rat-hours have no valid correlation and are left out. The integrated correlation is most negative from 17:00 to 21:00 (-0.37 to -0.51) and near zero from 02:00 to 05:00. Ten is fewer than the 13.8 pairs expected by chance among 276 uncorrected tests, so the difference from the paper is in the smallest p, not in a clear effect of hour; with this many uncorrected tests the criterion's bound is hard to meet whatever the data. The two alignments one reading apart give the same outcome (smallest p 0.032 and 0.015, with 2 and 13 pairs at p < 0.05).

How the test differed from the paper’s method

• Rats with at least 24 h of both signals, a standard choice; Spearman correlation as in edfig4-nrem-wake. A rat-hour where any of the four correlations cannot be computed is left out.

Interstitial glucose concentrations are modulated over the circadian timescale. f , Interstitial glucose concentrations increase during the postprandial period ( n = 36 meals, collected from 4 rats)

Results, SPW-Rs phase reset glucose oscillations, paragraph 1; Extended Data Fig. 1e-f

The meal times of the fasting experiment (36 meals, 4 rats) are not in the dandiset, and no subject is labelled as part of it, so the meal part and both expected values cannot be tested. The circadian part was run and does not find a modulation under the criterion's test (one-way ANOVA of hour of day on the hourly means of 25 subjects, F = 0.46, p = 0.99; 28.6 to 31.3 nA). That test pools hourly means across subjects whose offsets differ by tens of nA. Within subjects, with each run's linear decay and each subject's mean removed, hour of day is significant (one-way ANOVA p = 0.024, Friedman p = 5.2e-05; lowest at 06:00 to 07:00, -0.84 nA, highest at 13:00 to 14:00, +0.65 nA).

SPW-R events were only weakly correlated between septal and temporal poles of the hippocampus ... Ventral SPW-Rs were less strongly correlated with interstitial glucose fluctuations than were SPW-Rs recorded from the dorsal CA1 region

Results, Lateral septum relays SPW-Rs downstream, paragraph 1; Extended Data Fig. 9d-f

The dorsal and ventral rats are CGM37, CGM48 and Vanessa (tables Dorsalripples and Ventralripples). Their 5-min dorsal and ventral rates correlate with r = 0.44, 0.83 and 0.13. Only Vanessa has days with at least 4 h of glucose recorded with the electrophysiology (three, two of which give a correlation), too few for the criterion's t-tests; per day the ventral correlations at +10 min (-0.088, 0.096) are more negative than the dorsal (0.115, 0.239), paired p = 0.11, and Vanessa's sensor reads a median of 0.25 nA. CGM37 has no glucose recorded with the electrophysiology and CGM48 has 2.2 h, less than one day of at least 4 h, so the glucose the comparison needs is missing for two of the three rats. The rate-correlation part can be tested on its own (above); the claim's glucose part cannot.

In all cases, SPW-R rate was the most predictive variable

Results, SPW-Rs phase reset glucose oscillations, paragraph 3; Extended Data Fig. 6a-b; Supplementary Discussion, paragraph 4

Six of the seven predictors must be computed from the LFP (stored in chunks of 707 samples by 32 channels, so one channel means reading all) and the 5-kHz accelerometer over the 1717 h with both signals. The notebook times 30 min of each from 16 files, read in 4 processes (the cores of the machine running it) and in 16. The reading speed varies with the network; a single 30-min piece of LFP took from about 1 to 22 s in the last runs, and 106 s in an earlier one. In the final run, reading alone in 4 processes (4.4 s per hour of recording) would take about 127 min for all 18 rats, 108 min for the 8 rats with the most hours (the legend's n is 8) and 41 min for the 3 rats whose brain state is stored; in 16 processes (3.0 s per hour) about 85, 72 and 28 min, against the 30-min notebook limit. For the 3 rats the seven predictors and 100 splits of each model at 41 offsets come on top of the reading, and they are 3 of the paper's 8 rats; that test is an analysis not run in this attempt. Sleep states, the binarized brain state, are stored for 3 of the 18 rats; in the others the brain state would first have to be scored.