Sub-microsecond optical measurements of cell membrane charging and lesioning by pulsed electric fields (DANDI:001521)

Claims

From the paper Sub-microsecond optical measurements of cell membrane charging and lesioning by pulsed electric fields

Reproducing Semenov et al. (2026) from dandiset 001521 6

Notebook reproduce_semenov_2026.ipynb

Reproduced2

ReproducedResults, 3.6 Electroporative uptake of Ca2+, paragraph 2; Figure 6A
A single 1-us pulse causes clear Ca2+ entry into CHO cells at 1000 and 1200 V/cm, among field strengths escalated from 0 to 1200 V/cm.
Exposure to a single 1-μs pulse at escalating field strengths, from 0 to 1200 V/cm established a clear peak of Ca2+ entry at 1000 and 1200 V/cm
Results, 3.6 Electroporative uptake of Ca2+, paragraph 2; Figure 6A
Under M1 (1000 and 1200 V/cm at 250 and 300 V) three of the four readings hold at both fields. The end value (last 10 frames) is above the 0 V/cm group by 6.58 and 8.19% (Student's t-test p = 4.9e-09 and 3.0e-11), and the peak is above zero (2.30 and 5.35%, p = 3.3e-07 and 5.0e-07) and above the 0 V/cm peak (p = 6.1e-06 and 1.9e-08); the group means at both fields exceed those at 650 V/cm or less (0 and 150 V, at most -10.26% for the end value and 0.55% for the peak). The end value against zero fails, as it is negative at both fields (-5.44 and -3.83%), with the sham at -12.02%. Under M2 (300 and 350 V) the same three readings hold. With fields of view as the unit, the peak less the same field of view's sham is 2.42% at 250 V (4 fields of view, p = 0.0004) and 5.02% at 300 V (5, p = 0.0002). The end and peak readings compare later recordings with the sham, which is always recorded first, while the decline of the signal is steepest in the first recording (pre-pulse slope -3.97% per min in the first recording, -0.41 to -1.46% per min in the fifth to seventh). The step of dF at the pulse (5 frames after less 5 before), which this does not affect, is above the sham's at 250 V (0.35 +- 0.17%, p = 0.025) and 300 V (1.78 +- 0.31%, p = 3.2e-07), but not at 150 or 200 V (p = 0.13 and 0.10); at 250 V it is not above zero on its own (p = 0.058). So the entry at 1000 V/cm under M1 is small. The single-pulse session holds 27 cells in 6 fields of view, and three sets of fields of view give 14 cells, so the 14 cells of Fig. 6A cannot be identified. The results hold with regions of 0.5 and 1.0 of the radius and end windows of 5 and 20 frames.
How the test differed from the paper’s method
- The files give generator voltages and the paper field strengths, with no conversion in either. Two readings are run. M1, 3.9 V/cm per V, is adopted as the main reading (an assumption, as the paper gives no conversion), because the single-pulse voltages then fall on the paper's field series 570 to 1400 V/cm and the six train voltages nearest the fields labelled in Fig. 6B (230, 570, 650, 780, 900, 1000 V/cm) have group sizes close to the n the figure gives (28, 38, 36, 68, 47, 39 against 28, 37, 34, 54, 35, 36). M2, 3.3 V/cm per V, puts the train voltages 70, 200 and 230 V on the 230, 650 and 780 V/cm named in Section 3.6 and is shown as a variant. A named field is assigned to the nearest voltage. Neither factor follows from the electrode gap (V/d is 10 V/cm per V for the 1.00 mm of the files, 6.25 for the 1.6 mm of Methods 2.2). The outcome is the same under both.
- Cells are found by an Otsu threshold and watershed of the pre-pulse mean image, with a circular region of 0.7 of each cell's equivalent radius and the median of pixels more than 15 px from any cell as background; the paper drew the regions by hand (Methods 2.4) and these are standard choices, shown with regions of 0.5 and 1.0 of the radius, end windows of 5 and 20 frames, and a background taken as the 10th percentile of the cell-free pixels.
- Every detected cell in every file is used, as the dataset does not say which cells the paper used.
- The paper names no test for this panel; the claims file's readings (one-sample t-test or Student's t-test against 0 V/cm, end value or peak) are all run.
ReproducedResults, 3.6 Electroporative uptake of Ca2+, paragraph 2; Figure 6Aquestion to the reader
After a single 1-us pulse at 780 V/cm the Ca2+ signal rises only subtly, at the detection limit, and not significantly.
A subtle ΔF increase after exposure at 780 V/cm was at the detection limit and not statistically significant
Results, 3.6 Electroporative uptake of Ca2+, paragraph 2; Figure 6A
The criterion's test, a one-sample t-test of the end value against zero, finds no increase at 780 V/cm, with -7.42 +- 0.52% under M1 (200 V, n = 24) and -5.44 +- 0.55% under M2 (250 V, n = 18), one-sided p = 1 for both. That test cannot detect an increase here, as it finds none at 1000 and 1200 V/cm either (-5.44 and -3.83% under M1), so the outcome carries little information. Against the 0 V/cm group the end value is higher by 4.60% (M1, p = 1.3e-06), and the peak is 1.76 +- 0.61% (one-sample p = 0.0086; against the sham peak p = 0.021); both comparisons include the slowing of the decline with recording order. The step of dF at the pulse against the 0 V/cm group, which detects 1000 and 1200 V/cm under M1 (p = 0.025 and 3.2e-07), finds no increase at 780 V/cm under M1 (0.15 +- 0.16%, p = 0.10), and finds one under M2, where 780 V/cm is 250 V (p = 0.025). The step is the only reading not confounded by recording order, but it separates 1000 V/cm from the sham only barely (p = 0.025, and p = 0.058 against zero), and the steps at 780 and 1000 V/cm under M1 do not differ (0.15 +- 0.16 and 0.35 +- 0.17%, p = 0.40), so it is weak evidence for the absence at 780 V/cm. Whether the criterion's test can stand for the paper's is asked of a reader (questions/001521.yaml).
How the test differed from the paper’s method
- The files give generator voltages and the paper field strengths, with no conversion in either. Two readings are run. M1, 3.9 V/cm per V, is adopted as the main reading (an assumption, as the paper gives no conversion), because the single-pulse voltages then fall on the paper's field series 570 to 1400 V/cm and the six train voltages nearest the fields labelled in Fig. 6B (230, 570, 650, 780, 900, 1000 V/cm) have group sizes close to the n the figure gives (28, 38, 36, 68, 47, 39 against 28, 37, 34, 54, 35, 36). M2, 3.3 V/cm per V, puts the train voltages 70, 200 and 230 V on the 230, 650 and 780 V/cm named in Section 3.6 and is shown as a variant. A named field is assigned to the nearest voltage. Neither factor follows from the electrode gap (V/d is 10 V/cm per V for the 1.00 mm of the files, 6.25 for the 1.6 mm of Methods 2.2). The outcome is the same under both.
- Cells are found by an Otsu threshold and watershed of the pre-pulse mean image, with a circular region of 0.7 of each cell's equivalent radius and the median of pixels more than 15 px from any cell as background; the paper drew the regions by hand (Methods 2.4) and these are standard choices, shown with regions of 0.5 and 1.0 of the radius, end windows of 5 and 20 frames, and a background taken as the 10th percentile of the cell-free pixels.
- Every detected cell in every file is used, as the dataset does not say which cells the paper used.

Partially reproduced1

Partially reproducedResults, 3.5 Membrane charging in the presence of Gd3+ ions, paragraph 3; Figure 5B
By Eq. 2, a 10-fold increase of effective membrane conductivity from 0.0003 mS/m reduces the pole IMP by only about 2%, and a 100-fold increase by 12%.
its 10-fold increase would go unnoticed, yielding just a ~2% IMP reduction, and a 100-fold increase would reduce IMP by just 12%
Results, 3.5 Membrane charging in the presence of Gd3+ ions, paragraph 3; Figure 5B
Eq. 2 with the paper's constants gives f = 1.4801 for a 10-fold and 1.3220 for a 100-fold increase from 3e-7 S/m, a reduction of 1.33% of IMPC (outside 2 +- 0.2%) and 11.87% (within 12 +- 1.2%). Relative to f at the intact conductivity the reductions are 1.20 and 11.75%. This is a computation from the paper's equation and uses no recordings, so it says nothing about the dataset.

Not reproduced3

Not reproducedResults, 3.6 Electroporative uptake of Ca2+, paragraph 2; Figure 6A
After a single 1-us pulse at 1000 or 1200 V/cm the Ca2+ signal stays elevated through the 5-min observation period.
Ca2+ signal remained elevated through the 5-min observation period
Results, 3.6 Electroporative uptake of Ca2+, paragraph 2; Figure 6A
The mean dF over the last 10 frames is below zero at 1000 and 1200 V/cm under M1 (-5.44 +- 0.55% and -3.83 +- 0.71%) and at 1000 V/cm under M2 (-3.83%); at 1200 V/cm under M2 (350 V) it is +1.24 +- 1.95%, p = 0.53. The ratio of final to peak group mean is -63 and -2.4 under M1, -2.4 and 0.18 under M2. Unpulsed cells decline to -12.0% over the same window. The decline is in the cells' own fluorescence, -8.7% before background subtraction, while the cell-free background falls by 1.0%, so no choice of background region removes it. The paper's Fig. 6A shows the 0 V/cm mean trace rising to about +1% by 300 s, so these recordings or the processing behind the figure differ in a way the paper does not describe. Against the 0 V/cm group the end value is higher by 6.6 and 8.2% under M1 (p = 4.9e-09 and 3.0e-11), and less the same field of view's sham trace it is 6.4 and 8.4% above zero (p = 7.3e-10 and 4.5e-11), but these comparisons include the change of the decline with recording order. With a background taken as the 10th percentile of the cell-free pixels the sham still falls to -11.3% and the end values stay negative (-5.06 and -3.51%). The signal stays above that of unpulsed cells to the end of the 5 min, but not above its own pre-pulse level, which is what the criterion tests. With other region sizes and end windows the end values move by less than 1% and stay negative at 250 and 300 V. The question of how the paper's traces handle the decline is for the authors (see findings).
How the test differed from the paper’s method
- The files give generator voltages and the paper field strengths, with no conversion in either. Two readings are run. M1, 3.9 V/cm per V, is adopted as the main reading (an assumption, as the paper gives no conversion), because the single-pulse voltages then fall on the paper's field series 570 to 1400 V/cm and the six train voltages nearest the fields labelled in Fig. 6B (230, 570, 650, 780, 900, 1000 V/cm) have group sizes close to the n the figure gives (28, 38, 36, 68, 47, 39 against 28, 37, 34, 54, 35, 36). M2, 3.3 V/cm per V, puts the train voltages 70, 200 and 230 V on the 230, 650 and 780 V/cm named in Section 3.6 and is shown as a variant. A named field is assigned to the nearest voltage. Neither factor follows from the electrode gap (V/d is 10 V/cm per V for the 1.00 mm of the files, 6.25 for the 1.6 mm of Methods 2.2). The outcome is the same under both.
- Cells are found by an Otsu threshold and watershed of the pre-pulse mean image, with a circular region of 0.7 of each cell's equivalent radius and the median of pixels more than 15 px from any cell as background; the paper drew the regions by hand (Methods 2.4) and these are standard choices, shown with regions of 0.5 and 1.0 of the radius, end windows of 5 and 20 frames, and a background taken as the 10th percentile of the cell-free pixels.
- Every detected cell in every file is used, as the dataset does not say which cells the paper used.
Not reproducedResults, 3.6 Electroporative uptake of Ca2+, paragraph 3; Figure 6B
With trains of sixty 1-us pulses at 6 Hz (each cell tested once, 224 cells), a strong Ca2+ response occurs in 28% of cells at 780 V/cm and in nearly all cells at higher fields.
A strong Ca2+ response was recorded in 28% of cells exposed at 780 V/cm and in nearly all cells at higher intensities
Results, 3.6 Electroporative uptake of Ca2+, paragraph 3; Figure 6B
No reading meets the criterion. The fields above 780 V/cm are taken as those of Fig. 6B, 900 and 1000 V/cm (230 and 250 V under M1, 250 and 300 V under M2). The fraction of strong responders at 780 V/cm is 2.9 to 22.1% under M1 (200 V, 68 cell exposures) and 44.7 to 78.7% under M2 (230 V, 47), against 28 +- 2.8%, over six rules, including "above 3 s.d." read literally as a threshold of 3 s.d. of the reference group. The lower of the two fractions at the higher fields is 44.7 to 78.7% under M1 and 61.5 to 79.5% under M2, against at least 90%. With regions of 0.5 and 1.0 of the radius the criterion is met in none of the 36 combinations of region, conversion and rule. The six voltage groups that correspond to the Fig. 6B fields hold 256 cell exposures under M1 (224 under M2), against 224; in all, the dataset holds 311 exposures to trains of 82 distinct cells in 9 fields of view, each recorded at 1 to 7 voltages, where the paper states 224 cells each tested once (measured gives the 780 V/cm fraction under M1 and rule (a) against 0 V/cm, and the cells in the M1 groups of the Fig. 6B fields). The figure's group sizes are close to the dataset's exposures, so the paper may have counted exposures as cells. Taking only the first train of each field of view, 9.1 to 18.2% respond at 200 V and 90.9 to 100% at 250 V (11 cells each). The end values carry the decline of the sham signal (-15.7% in the train session) that the Fig. 6B traces do not show. These numbers use the criterion's end window of 10 frames; with 5 frames and a region of 0.5 of the radius the 780 V/cm fraction under M1 and the rule above the largest of the lowest field is 27.9%, within tolerance, but the higher fields reach 76.6 and 82.1%, not 90%. With a background taken as the 10th percentile of the cell-free pixels, which dim unsegmented cells affect less (the background falls from 1.81 to 1.21 times the corrected cell signal), the criterion is met in none of 12 combinations of conversion and rule (780 V/cm 2.9 to 23.5% under M1, 42.6 to 91.5% under M2; lower higher-field fraction at most 87.2%).
How the test differed from the paper’s method
- The files give generator voltages and the paper field strengths, with no conversion in either. Two readings are run. M1, 3.9 V/cm per V, is adopted as the main reading (an assumption, as the paper gives no conversion), because the single-pulse voltages then fall on the paper's field series 570 to 1400 V/cm and the six train voltages nearest the fields labelled in Fig. 6B (230, 570, 650, 780, 900, 1000 V/cm) have group sizes close to the n the figure gives (28, 38, 36, 68, 47, 39 against 28, 37, 34, 54, 35, 36). M2, 3.3 V/cm per V, puts the train voltages 70, 200 and 230 V on the 230, 650 and 780 V/cm named in Section 3.6 and is shown as a variant. A named field is assigned to the nearest voltage. Neither factor follows from the electrode gap (V/d is 10 V/cm per V for the 1.00 mm of the files, 6.25 for the 1.6 mm of Methods 2.2). The outcome is the same under both.
- Cells are found by an Otsu threshold and watershed of the pre-pulse mean image, with a circular region of 0.7 of each cell's equivalent radius and the median of pixels more than 15 px from any cell as background; the paper drew the regions by hand (Methods 2.4) and these are standard choices, shown with regions of 0.5 and 1.0 of the radius, end windows of 5 and 20 frames, and a background taken as the 10th percentile of the cell-free pixels.
- The cells were tested more than once in the dataset; all exposures are used, and the first exposure of each field of view is shown as a variant.
- Reading (b), above 3 s.d. of the reference group, is run as above its mean plus 3 s.d. and, literally, above 3 s.d.
Not reproducedResults, 3.6 Electroporative uptake of Ca2+, paragraph 3; Figure 6Bverifier disagreed at first
With 60-pulse trains at 650 V/cm only 2 cells (5.9%) respond strongly, but the mean Ca2+ signal at the end of recording is significantly above zero and marginally above that at 230 V/cm.
At 650 V/cm, only 2 cells (5.9%) showed a strong response, but the mean Ca2+ signal by the end of recording was significantly different from zero (p < 0.001, one sample t-test) and marginally higher than after exposure at 230 V/cm (p < 0.05, single-tailed t-test)
Results, 3.6 Electroporative uptake of Ca2+, paragraph 3; Figure 6B
What the paper reports
- 650 V/cm end-of-recording signal against zero: one sample t-test, p < 0.001
- 650 V/cm against 230 V/cm: single-tailed t-test, p < 0.05
The statement's tests fail under both conversions. Under M1 the mean end value at 170 V (650 V/cm) is -9.17 +- 0.83%, significantly below zero (two-sided p = 6.4e-13, one-sided p = 1 for above zero), and not above the 70 V group (single-tailed p = 0.20). Under M2 the 200 V mean is -1.98 +- 1.77% (p = 0.27 against zero) and above the 70 V group (single-tailed p = 0.0023). The count is met under one combination of 36 (M1, 170 V, rule above the largest of the lowest field, region 0.7 of the radius, 2 of 36, 5.6%); measured gives the first reading, rule (a) against 0 V/cm under M1, 0 cells. Regions of 0.5 and 1.0 give 1 cell (2.8%) under the same rule, the other five rules give 0 under M1, and M2 (200 V, 68 exposures) gives 2 to 17 cells (2.9 to 25.0%). As the tests fail under both conversions and every rule, the outcome does not rest on the choice. The end values carry the decline of the sham signal. With a background taken as the 10th percentile of the cell-free pixels the 650 V/cm mean is -7.37% under M1 and -1.54% under M2 (one-sided p = 1.0 and 0.87 for above zero), and the count of 2 cells (5.9%) is met in none of 12 combinations. In round 1 this claim was blocked as methods_underspecified; the statement's tests are fully specified and fail, so it is not reproduced.
How the test differed from the paper’s method
- The files give generator voltages and the paper field strengths, with no conversion in either. Two readings are run. M1, 3.9 V/cm per V, is adopted as the main reading (an assumption, as the paper gives no conversion), because the single-pulse voltages then fall on the paper's field series 570 to 1400 V/cm and the six train voltages nearest the fields labelled in Fig. 6B (230, 570, 650, 780, 900, 1000 V/cm) have group sizes close to the n the figure gives (28, 38, 36, 68, 47, 39 against 28, 37, 34, 54, 35, 36). M2, 3.3 V/cm per V, puts the train voltages 70, 200 and 230 V on the 230, 650 and 780 V/cm named in Section 3.6 and is shown as a variant. A named field is assigned to the nearest voltage. Neither factor follows from the electrode gap (V/d is 10 V/cm per V for the 1.00 mm of the files, 6.25 for the 1.6 mm of Methods 2.2). The outcome is the same under both.
- Cells are found by an Otsu threshold and watershed of the pre-pulse mean image, with a circular region of 0.7 of each cell's equivalent radius and the median of pixels more than 15 px from any cell as background; the paper drew the regions by hand (Methods 2.4) and these are standard choices, shown with regions of 0.5 and 1.0 of the radius, end windows of 5 and 20 frames, and a background taken as the 10th percentile of the cell-free pixels.

Not tested in a notebook 21

21 data missing from the dataset

Blocked21

BlockedResults, 3.1 Visualization of electroporation, paragraph 1; Figure 1· Data missing from the dataset
With Di-8-ANEPPS, polarization by 1-us pulses grows with field strength up to 570 to 780 V/cm; at stronger fields the poles become leaky and show reduced or absent polarization while regions near the equator stay polarized.
Still stronger electric fields produce a distinct pattern of reduced or absent polarization at the poles (where the membrane becomes too leaky to accumulate the charge) while regions closer to the equator remain relatively intact and may polarize more than the poles
Results, 3.1 Visualization of electroporation, paragraph 1; Figure 1
The pulsed-laser potentiometric imaging (FluoVolt and Di-8-ANEPPS) is not in this dandiset, which holds only Fluo-4 time-lapses. The paper's Data availability says data are available on request. Six other dandisets of the same lab name this paper in their metadata (001515, 001516, 001517, 001518, 001525 and 001526) and by their titles hold the potentiometric experiments; they are targets of their own, and the claims file does not name them as data the paper relies on, so they were not used here.
BlockedResults, 3.2 Polarization by multiple repeated stimuli, paragraph 2; Figure 2A· Data missing from the dataset
FluoVolt dF rises with field strength until it saturates at 66% (255 mV) at 360 V/cm at the cathodic pole and at -83% (-320 mV) at 570 V/cm at the anodic pole; stronger fields do not increase it at the anodic pole and decrease it at the cathodic pole.
ΔF increased proportionally to the electric field strength until reaching apparent saturation at 66% ΔF (255 mV IMP) at 360 V/cm at the cathodic pole, and −83% ΔF (−320 mV IMP) at 570 V/cm at the anodic pole
Results, 3.2 Polarization by multiple repeated stimuli, paragraph 2; Figure 2A
The pulsed-laser potentiometric imaging (FluoVolt and Di-8-ANEPPS) is not in this dandiset, which holds only Fluo-4 time-lapses. The paper's Data availability says data are available on request. Six other dandisets of the same lab name this paper in their metadata (001515, 001516, 001517, 001518, 001525 and 001526) and by their titles hold the potentiometric experiments; they are targets of their own, and the claims file does not name them as data the paper relies on, so they were not used here.
BlockedResults, 3.2 Polarization by multiple repeated stimuli, paragraph 2; Figure 2A· Data missing from the dataset
FluoVolt responses stayed stable through the 60 pulses of each set, even at 1000 V/cm and above, which electroporate strongly.
Field strengths of 1000 V/cm and higher caused strong electroporation (see Section 3.5), but the dye response in each set stayed stable through the 60 pulses applied
Results, 3.2 Polarization by multiple repeated stimuli, paragraph 2; Figure 2A
The pulsed-laser potentiometric imaging (FluoVolt and Di-8-ANEPPS) is not in this dandiset, which holds only Fluo-4 time-lapses. The paper's Data availability says data are available on request. Six other dandisets of the same lab name this paper in their metadata (001515, 001516, 001517, 001518, 001525 and 001526) and by their titles hold the potentiometric experiments; they are targets of their own, and the claims file does not name them as data the paper relies on, so they were not used here.
BlockedResults, 3.2 Polarization by multiple repeated stimuli, paragraph 3; Figure 2B· Data missing from the dataset
Di-8-ANEPPS dF keeps rising up to 570 V/cm at both poles, and at 780 V/cm and above it declines gradually within each set of 60 pulses.
In addition, responses to PEFs at and above 780 V/cm decreased gradually in each set of 60 images
Results, 3.2 Polarization by multiple repeated stimuli, paragraph 3; Figure 2B
The pulsed-laser potentiometric imaging (FluoVolt and Di-8-ANEPPS) is not in this dandiset, which holds only Fluo-4 time-lapses. The paper's Data availability says data are available on request. Six other dandisets of the same lab name this paper in their metadata (001515, 001516, 001517, 001518, 001525 and 001526) and by their titles hold the potentiometric experiments; they are targets of their own, and the claims file does not name them as data the paper relies on, so they were not used here.
BlockedResults, 3.2 Polarization by multiple repeated stimuli, paragraph 4; Figure 2C and 2D· Data missing from the dataset
Alternating sets at 150 and 230 V/cm give dF that stays constant within and between sets; after electroporation at 1000 V/cm the set-averaged dF drops by about 20% and stays lower.
ΔF values averaged across each set of 60 PEFs stayed nearly constant before the electroporation, but dropped by about 20% in the sets acquired after it (Fig. 2D, p < 0.05)
Results, 3.2 Polarization by multiple repeated stimuli, paragraph 4; Figure 2C and 2D
What the paper reports
- 2-tailed paired t-test, p < 0.05 (n = 8 cells, one per experiment)
The pulsed-laser potentiometric imaging (FluoVolt and Di-8-ANEPPS) is not in this dandiset, which holds only Fluo-4 time-lapses. The paper's Data availability says data are available on request. Six other dandisets of the same lab name this paper in their metadata (001515, 001516, 001517, 001518, 001525 and 001526) and by their titles hold the potentiometric experiments; they are targets of their own, and the claims file does not name them as data the paper relies on, so they were not used here.
BlockedResults, 3.2 Polarization by multiple repeated stimuli, paragraph 4; Figure 2C· Data missing from the dataset
At 1000 V/cm, cathodic dF fell from 40% at the start of the set to 26.6% at its end, a trend that was not significant.
ΔF at the cathodic pole decreased from 40 ± 10% at the beginning of the set to 26.6 ± 8% at the end of it. Although not statistically significant (p > 0.05)
Results, 3.2 Polarization by multiple repeated stimuli, paragraph 4; Figure 2C
What the paper reports
- p > 0.05 (n = 8 cells)
The pulsed-laser potentiometric imaging (FluoVolt and Di-8-ANEPPS) is not in this dandiset, which holds only Fluo-4 time-lapses. The paper's Data availability says data are available on request. Six other dandisets of the same lab name this paper in their metadata (001515, 001516, 001517, 001518, 001525 and 001526) and by their titles hold the potentiometric experiments; they are targets of their own, and the claims file does not name them as data the paper relies on, so they were not used here.
BlockedResults, 3.3 Calibration of the potentiometric dyes, paragraph 1; Figure 3· Data missing from the dataset
Fitted through the origin over -150 to +150 mV of calculated IMP, FluoVolt and Di-8-ANEPPS change emission by 26% and 9.1% per 100 mV, with correlation coefficients of 0.96 and 0.78.
The correlation coefficients of the fits equaled 0.96 for FluoVolt and 0.78 for Di-8 ANEPPS (due to a lower signal-to-noise ratio), with respective slope coefficients of 26% and 9.1% ΔF per 100 mV of IMP
Results, 3.3 Calibration of the potentiometric dyes, paragraph 1; Figure 3
The pulsed-laser potentiometric imaging (FluoVolt and Di-8-ANEPPS) is not in this dandiset, which holds only Fluo-4 time-lapses. The paper's Data availability says data are available on request. Six other dandisets of the same lab name this paper in their metadata (001515, 001516, 001517, 001518, 001525 and 001526) and by their titles hold the potentiometric experiments; they are targets of their own, and the claims file does not name them as data the paper relies on, so they were not used here.
BlockedResults, 3.3 Calibration of the potentiometric dyes, paragraph 1; Figure 3· Data missing from the dataset
FluoVolt becomes non-linear already at plus or minus 200 mV while Di-8-ANEPPS stays nearly linear from -500 to +500 mV, so the early departure of FluoVolt reflects its dynamic range, not electroporation.
The actual measurements with FluoVolt gradually became non-linear already at ±200 mV, while those for Di-8-ANEPPS stayed essentially linear throughout the tested range from −500 to +500 mV
Results, 3.3 Calibration of the potentiometric dyes, paragraph 1; Figure 3
The pulsed-laser potentiometric imaging (FluoVolt and Di-8-ANEPPS) is not in this dandiset, which holds only Fluo-4 time-lapses. The paper's Data availability says data are available on request. Six other dandisets of the same lab name this paper in their metadata (001515, 001516, 001517, 001518, 001525 and 001526) and by their titles hold the potentiometric experiments; they are targets of their own, and the claims file does not name them as data the paper relies on, so they were not used here. It also depends on fig3-dye-calibration, blocked for the same reason.
BlockedResults, 3.4 Azimuthal membrane charging and polarization limits, paragraph 1; Figure 4A· Data missing from the dataset
With FluoVolt the maximum dF reached 71% and -83.5% at the cathodic and anodic poles, and cathodic polarization fell from 71% at 360 V/cm to 45 to 50% at 1000 to 1400 V/cm.
With stronger PEFs, maximum ΔF reached 71 ± 10% and −83.5 ± 5% at the cathodic and anodic poles, respectively ... polarization at the cathodic pole decreased from 71% at 360 V/cm to 45–50% at 1000–1400 V/cm
Results, 3.4 Azimuthal membrane charging and polarization limits, paragraph 1; Figure 4A
The pulsed-laser potentiometric imaging (FluoVolt and Di-8-ANEPPS) is not in this dandiset, which holds only Fluo-4 time-lapses. The paper's Data availability says data are available on request. Six other dandisets of the same lab name this paper in their metadata (001515, 001516, 001517, 001518, 001525 and 001526) and by their titles hold the potentiometric experiments; they are targets of their own, and the claims file does not name them as data the paper relies on, so they were not used here.
BlockedResults, 3.4 Azimuthal membrane charging and polarization limits, paragraph 2; Figure 4B· Data missing from the dataset
With Di-8-ANEPPS, measured IMP fell below the calculated value at 570 V/cm, where dF reached its maximum of 57% at the cathodic and -55% at the anodic pole, over 600 mV, twice the FluoVolt value.
The measured IMP decreased below the calculated value when the electric field reached 570 V/cm, concurrently with reaching the maximum ΔF of 57 ± 14% at the cathodic pole and −55 ± 4% at the anodic pole
Results, 3.4 Azimuthal membrane charging and polarization limits, paragraph 2; Figure 4B
The pulsed-laser potentiometric imaging (FluoVolt and Di-8-ANEPPS) is not in this dandiset, which holds only Fluo-4 time-lapses. The paper's Data availability says data are available on request. Six other dandisets of the same lab name this paper in their metadata (001515, 001516, 001517, 001518, 001525 and 001526) and by their titles hold the potentiometric experiments; they are targets of their own, and the claims file does not name them as data the paper relies on, so they were not used here.
BlockedResults, 3.4 Azimuthal membrane charging and polarization limits, paragraph 3; Figure 4B (dotted lines)· Data missing from the dataset
Averaged over the first 10 of 60 pulses, Di-8-ANEPPS polarization at the cathodic pole reached 66% (710 mV).
Polarization measured this way reached 66 ± 15% (710 ± 200 mV) at the cathodic pole
Results, 3.4 Azimuthal membrane charging and polarization limits, paragraph 3; Figure 4B (dotted lines)
The pulsed-laser potentiometric imaging (FluoVolt and Di-8-ANEPPS) is not in this dandiset, which holds only Fluo-4 time-lapses. The paper's Data availability says data are available on request. Six other dandisets of the same lab name this paper in their metadata (001515, 001516, 001517, 001518, 001525 and 001526) and by their titles hold the potentiometric experiments; they are targets of their own, and the claims file does not name them as data the paper relies on, so they were not used here.
BlockedResults, 3.4 Azimuthal membrane charging and polarization limits, paragraph 4; Figure 4B· Data missing from the dataset
Above 780 V/cm the IMP at the poles falls progressively with field strength, more at the cathodic pole; at 1400 V/cm charging is undetectable at the cathodic pole and about -30% dF at the anodic pole.
As the electric field increased further, the IMP reached at the poles progressively decreased, and the reduction was more pronounced at the cathodic pole ... At 1400 V/cm, charging was no longer detectable at the cathodic pole
Results, 3.4 Azimuthal membrane charging and polarization limits, paragraph 4; Figure 4B
The pulsed-laser potentiometric imaging (FluoVolt and Di-8-ANEPPS) is not in this dandiset, which holds only Fluo-4 time-lapses. The paper's Data availability says data are available on request. Six other dandisets of the same lab name this paper in their metadata (001515, 001516, 001517, 001518, 001525 and 001526) and by their titles hold the potentiometric experiments; they are targets of their own, and the claims file does not name them as data the paper relies on, so they were not used here.
BlockedResults, 3.4 Azimuthal membrane charging and polarization limits, paragraph 4; Figure 4C· Data missing from the dataset
Membrane regions near the equator are relatively unaffected and follow the cosine dependence of the Schwan equation up to 1000 V/cm.
Membrane areas closer to the equator (at polar angles closer to 0◦and 270◦) appeared relatively unaffected and followed the cosine dependence predicted by Schwan formula up to 1000 V/cm
Results, 3.4 Azimuthal membrane charging and polarization limits, paragraph 4; Figure 4C
The pulsed-laser potentiometric imaging (FluoVolt and Di-8-ANEPPS) is not in this dandiset, which holds only Fluo-4 time-lapses. The paper's Data availability says data are available on request. Six other dandisets of the same lab name this paper in their metadata (001515, 001516, 001517, 001518, 001525 and 001526) and by their titles hold the potentiometric experiments; they are targets of their own, and the claims file does not name them as data the paper relies on, so they were not used here.
BlockedResults, 3.4 Azimuthal membrane charging and polarization limits, paragraph 5; Figure 4D· Data missing from the dataset
Shortening the pulse to 0.3 us at higher fields (357 to 3320 V/cm) did not raise the polarization limit; the maximum dF was 53% (about 600 mV) at both poles, as with 1-us pulses.
Contrary to the expectations, the results were similar to those with 1-μs pulses in Fig. 4B ... the maximum ΔF reached was 53 ± 4% (~600 mV) at both cell poles
Results, 3.4 Azimuthal membrane charging and polarization limits, paragraph 5; Figure 4D
The pulsed-laser potentiometric imaging (FluoVolt and Di-8-ANEPPS) is not in this dandiset, which holds only Fluo-4 time-lapses. The paper's Data availability says data are available on request. Six other dandisets of the same lab name this paper in their metadata (001515, 001516, 001517, 001518, 001525 and 001526) and by their titles hold the potentiometric experiments; they are targets of their own, and the claims file does not name them as data the paper relies on, so they were not used here. It also depends on fig4b-di8-maximum, blocked for the same reason.
BlockedResults, 3.5 Membrane charging in the presence of Gd3+ ions, paragraph 1; Figure 5A· Data missing from the dataset
1 mM Gd3+ extended anodic charging to -864 mV against -542 mV in parallel controls, and the measured IMP stayed close to the calculated IMP down to -780 mV.
Indeed, we found that 1 mM Gd3+ extended the range of membrane charging at the anodic pole down to −864 ± 44 mV (p < 0.001 compared to −542 ± 57 mV in parallel control experiments, Fig. 5A)
Results, 3.5 Membrane charging in the presence of Gd3+ ions, paragraph 1; Figure 5A
What the paper reports
- 2-tailed t-test, p < 0.001 (n = 7 cells with Gd3+ and 5 control cells)
The pulsed-laser potentiometric imaging (FluoVolt and Di-8-ANEPPS) is not in this dandiset, which holds only Fluo-4 time-lapses. The paper's Data availability says data are available on request. Six other dandisets of the same lab name this paper in their metadata (001515, 001516, 001517, 001518, 001525 and 001526) and by their titles hold the potentiometric experiments; they are targets of their own, and the claims file does not name them as data the paper relies on, so they were not used here.
BlockedResults, 3.5 Membrane charging in the presence of Gd3+ ions, paragraph 2; Figure 5A· Data missing from the dataset
At the cathodic pole the IMP with Gd3+ reached 647 mV, not significantly higher than in controls, but Gd3+ prevented the decline of IMP with stronger fields.
At the cathodic pole, polarization in the presence of Gd3+ reached 647 ± 44 mV, a value not significantly higher than that in the control group without Gd3+. However, Gd3+ prevented the progressive decline in IMP as the electric field was increased
Results, 3.5 Membrane charging in the presence of Gd3+ ions, paragraph 2; Figure 5A
The pulsed-laser potentiometric imaging (FluoVolt and Di-8-ANEPPS) is not in this dandiset, which holds only Fluo-4 time-lapses. The paper's Data availability says data are available on request. Six other dandisets of the same lab name this paper in their metadata (001515, 001516, 001517, 001518, 001525 and 001526) and by their titles hold the potentiometric experiments; they are targets of their own, and the claims file does not name them as data the paper relies on, so they were not used here.
BlockedResults, 3.5 Membrane charging in the presence of Gd3+ ions, paragraph 4; Figure 5C· Data missing from the dataset
Gd3+ reduced the electroporation-induced effective membrane conductivity 2 to 3-fold for most datapoints.
For most datapoints, Gd3+ reduced the electroporation- induced effective conductivity 2–3-fold
Results, 3.5 Membrane charging in the presence of Gd3+ ions, paragraph 4; Figure 5C
The pulsed-laser potentiometric imaging (FluoVolt and Di-8-ANEPPS) is not in this dandiset, which holds only Fluo-4 time-lapses. The paper's Data availability says data are available on request. Six other dandisets of the same lab name this paper in their metadata (001515, 001516, 001517, 001518, 001525 and 001526) and by their titles hold the potentiometric experiments; they are targets of their own, and the claims file does not name them as data the paper relies on, so they were not used here. It also needs the pole IMP of Fig. 5A; its dependency fig5b-conductivity-calibration was tested.
BlockedResults, 3.7 Fast restoration of membrane conductivity, paragraph 3; Figure 7A· Data missing from the dataset
A 150 V/cm test pulse 5 us after a conditioning pulse still charges the membrane along a cosine after conditioning up to 1000 V/cm, while 1200 and 1400 V/cm conditioning markedly hinders charging at the poles.
Polarization still followed a cosine function after conditioning PEFs of up to 1000 V/cm. However, 1200 and 1400 V/cm conditioning PEFs markedly hindered charging by the test pulse at cell poles
Results, 3.7 Fast restoration of membrane conductivity, paragraph 3; Figure 7A
The pulsed-laser potentiometric imaging (FluoVolt and Di-8-ANEPPS) is not in this dandiset, which holds only Fluo-4 time-lapses. The paper's Data availability says data are available on request. Six other dandisets of the same lab name this paper in their metadata (001515, 001516, 001517, 001518, 001525 and 001526) and by their titles hold the potentiometric experiments; they are targets of their own, and the claims file does not name them as data the paper relies on, so they were not used here. It also depends on fig7b-test-pulse-recovery, blocked for the same reason.
BlockedResults, 3.7 Fast restoration of membrane conductivity, paragraph 5; Figure 7B top· Data missing from the dataset
During the conditioning pulse, pole dF falls 1.5 times below linear scaling at 560 V/cm, 2-fold at 760 V/cm and further at stronger fields.
Actual measurements fell 1.5 times below this line for 560 V/cm PEFs, 2- fold for 760 V/cm PEFs, and diverged further for more intense PEFs (p < 0.01 for all datapoints ≥560 V/cm, one sample t-test)
Results, 3.7 Fast restoration of membrane conductivity, paragraph 5; Figure 7B top
What the paper reports
- one sample t-test, p < 0.01 (n = 5 cells)
The pulsed-laser potentiometric imaging (FluoVolt and Di-8-ANEPPS) is not in this dandiset, which holds only Fluo-4 time-lapses. The paper's Data availability says data are available on request. Six other dandisets of the same lab name this paper in their metadata (001515, 001516, 001517, 001518, 001525 and 001526) and by their titles hold the potentiometric experiments; they are targets of their own, and the claims file does not name them as data the paper relies on, so they were not used here.
BlockedResults, 3.7 Fast restoration of membrane conductivity, paragraph 5; Figure 7B bottom· Data missing from the dataset
Six us after the conditioning pulse, test-pulse charging is unchanged after conditioning up to 760 V/cm, about 80% (not significant) after 1000 V/cm, and about 70% and 50% after 1200 and 1400 V/cm.
Still more intense PEFs of 1200 and 1400 V/cm reduced charging by the test pulse to ~70% and ~ 50%, respectively (p < 0.05 and p < 0.01, two-sided t-test)
Results, 3.7 Fast restoration of membrane conductivity, paragraph 5; Figure 7B bottom
What the paper reports
- 1200 V/cm conditioning: two-sided t-test, p < 0.05 (n = 5 cells)
- 1400 V/cm conditioning: two-sided t-test, p < 0.01 (n = 5 cells)
The pulsed-laser potentiometric imaging (FluoVolt and Di-8-ANEPPS) is not in this dandiset, which holds only Fluo-4 time-lapses. The paper's Data availability says data are available on request. Six other dandisets of the same lab name this paper in their metadata (001515, 001516, 001517, 001518, 001525 and 001526) and by their titles hold the potentiometric experiments; they are targets of their own, and the claims file does not name them as data the paper relies on, so they were not used here.
BlockedResults, 3.7 Fast restoration of membrane conductivity, paragraph 7; Figure 7C· Data missing from the dataset
Effective membrane conductivity at the poles dropped about 10-fold (range 7 to 20) within 6 us after the electroporating pulse, with full resealing after 570 and 760 V/cm and residual conductivity after stronger pulses.
Fig. 7C shows a profound reduction in the λm,eff within just 6 μs after the electroporating pulse. For most datapoints, it dropped by a factor of 10 (range ... 7–20)
Results, 3.7 Fast restoration of membrane conductivity, paragraph 7; Figure 7C
The pulsed-laser potentiometric imaging (FluoVolt and Di-8-ANEPPS) is not in this dandiset, which holds only Fluo-4 time-lapses. The paper's Data availability says data are available on request. Six other dandisets of the same lab name this paper in their metadata (001515, 001516, 001517, 001518, 001525 and 001526) and by their titles hold the potentiometric experiments; they are targets of their own, and the claims file does not name them as data the paper relies on, so they were not used here. It also depends on fig7b-test-pulse-recovery, blocked for the same reason.

Findings about the data

What attempts found about the dataset itself, for its owner and the archive.

Data missingThe dandiset holds only the Fluo-4 imaging of Fig. 6. The pulsed-laser potentiometric imaging behind Figs. 1 to 5 and 7, which carries most of the paper's findings (21 of 27 claims), is not in it. Six other dandisets of the lab (001515, 001516, 001517, 001518, 001525, 001526) name the same paper and by their titles hold these experiments, but neither the paper's Data availability (on request) nor this dandiset's description points to them.
Description missingThe files give each pulse as a generator voltage (0 to 400 V); the paper reports only field strengths and gives no conversion. At about 3.9 V/cm per V the single-pulse voltages fall on the paper's field series and the six train voltages nearest the Fig. 6B fields have group sizes close to the figure's n (28, 38, 36, 68, 47, 39 against 28, 37, 34, 54, 35, 36). At about 3.3 V/cm per V the train voltages fall on the 230, 650 and 780 V/cm named in the text. The conversion is an assumption here.
Metadata errorEvery file states an electrode gap of 1.00 mm and an electrode diameter of 0.5 mm; the paper's Methods 2.2 give tungsten rods 500 um in diameter at 1.6 mm spacing.
OtherIn the train session (2025-05-02) each subject is one field of view recorded at 1 to 7 escalating voltages after a sham, so the same 82 cells account for 311 exposures to trains. The paper states that each cell was tested only once, 28 to 54 cells per field strength, 224 cells in all. Either the dandiset holds different recordings from Fig. 6B, or the paper's description is wrong.
Selection not documentedThe single-pulse session holds 27 cells in 6 fields of view, recorded at 0 to 400 V; Fig. 6A averages 14 cells over 0 to 1200 V/cm. Three different sets of fields of view hold 14 cells. The dandiset does not say which cells or recordings the paper used, and it holds no cell regions or traces, so the paper's selection cannot be repeated.
Metadata errorThe dandiset's description and title describe only single 1-us pulses with recordings of 5 min at 1 frame/s, but 33 of the 73 files are 60-pulse 6 Hz trains recorded for 3 min at 0.5 frame/s (the other 9 files of that session are shams). Subjects are fields of view of a cell line, yet each has an age of P1D, and the archive's species summary names two species, 'Cricetulus griseus - Cricetulus aureus', for the one taxon NCBITaxon 10029.
Methods not availableQuestion for the authors. How were generator voltages converted to the field strengths of Fig. 6, and was the same conversion used for single pulses and trains? 3.9 V/cm per V fits the fields and group sizes of Fig. 6B and was adopted; 3.3 V/cm per V puts the train voltages on the three fields named in the text and was run as a variant. No outcome changes between the two.
Methods not availableQuestion for the authors. How was a strong Ca2+ response defined in Fig. 6B, how large was the region over each cell, and which cells formed each group? The count of 2 responders at 650 V/cm is met under one of 36 combinations of conversion, rule and region size here.
Methods not availableQuestion for the authors. Unpulsed cells lose 12% (single-pulse session) and 16% (train session) of their background-corrected Fluo-4 signal by the end of a recording, relative to the pre-pulse mean; the raw cell signal falls by 8.7 and 7.5% while the cell-free background falls by 1.0 and 0.6%. The 0 V/cm trace of Fig. 6A rises by about 1% and the Fig. 6B traces are flat. Were the dF traces of Fig. 6 corrected for this decline, or do they come from other recordings, and which test gave "not statistically significant" at 780 V/cm? Without a correction the end-of-recording dF at 1000 and 1200 V/cm is below zero, and fig6a-ca-sustained is not reproduced.
Paper inconsistentEq. 2 with the constants of Methods 2.3 gives a 1.3% reduction of pole IMP for a 10-fold increase of membrane conductivity, where Section 3.5 states about 2%.
OtherEach field of view was recorded first without a pulse and then at escalating voltages, and the decline of the Fluo-4 signal is steepest in the first recording (before the pulse, -3.97% per min in the first recording of the single-pulse session and -0.41 to -1.46% per min in the fifth to seventh, slowing with each recording; -11.4% per min in the first recording of the train session and -2.0 to -5.4% per min in later ones, without a steady trend). Field strength and recording order therefore go together, and comparisons with the sham, always recorded first, include part of this difference.

Questions

Where the author or a verifier believed the claims file misread the paper, they asked, citing the paper, and a reader who had not seen the data answered from the paper alone, amending the claims file where it was wrong. The author then ran the claim as answered. Attempts before protocol pilot-11 also asked here for method details the claims file did not give.

openasked 2026-11-03about fig6a-780-not-significant
The criterion judges the claim of absence by a one-sample t-test of each cell's end-of-recording dF against zero. Can that test stand for the paper's "not statistically significant", when the same claims file treats 1000 and 1200 V/cm as a clear entry? In this dataset that test finds no increase at 1000 and 1200 V/cm either, so it cannot detect the effect the claim says is absent at 780 V/cm. Should the test be the one that detects the entry at 1000 and 1200 V/cm (for example against the 0 V/cm group, or on the change at the pulse), with the claim of absence judged against that?
Passage. Exposure to a single 1-us pulse at escalating field strengths, from 0 to 1200 V/cm established a clear peak of Ca2+ entry at 1000 and 1200 V/cm, and Ca2+ signal remained elevated through the 5-min observation period (Fig. 6A). A subtle dF increase after exposure at 780 V/cm was at the detection limit and not statistically significant. (Section 3.6, paragraph 2; Methods 2.5 name Student's t-test and the one-sample t-test.)
Why it matters. Unpulsed cells decline to -12% by the end of the recording here, so the end value is below zero at 780, 1000 and 1200 V/cm (under 3.9 V/cm per V, -7.42, -5.44 and -3.83%) and the criterion's test gives "reproduced" without being able to find any increase. Against the 0 V/cm group the end value and the peak are significantly higher at 780 V/cm (p = 1.3e-06 and 0.021), which would give "not reproduced"; the step of dF at the pulse against the 0 V/cm group detects 1000 and 1200 V/cm (p = 0.025 and 3e-07) and not 780 V/cm (p = 0.10), which would give "reproduced". The verifier of round 1 reported the criterion's test as unable to detect an increase. The verifier of round 2 noted that the step reading is marginal at 1000 V/cm (p = 0.058 against zero) and that the steps at 780 and 1000 V/cm were not shown to differ; they do not (p = 0.40), so no reading here both detects 1000 V/cm clearly and finds 780 V/cm absent.
No reader has answered it yet.

Notebooks

Every registered entry that uses this dataset. An entry can hold several notebooks, and can use more than one dataset.

Reproduction Ca2+ uptake after 1-us pulses in CHO cells (Semenov et al. 2026)

Tests the findings of Semenov, Silkuniene, Silkunas, Pakhomova, Bixler, Ibey, Kochunov and Pakhomov (2026) that can be tested on dandiset 001521, which holds the Fluo-4 imaging of CHO cells after single 1-us pulses and after 60-pulse trains (Fig. 6). It segments the cells, computes dF for each, and compares field strengths under two readings of the voltage to field conversion. The potentiometric imaging of Figures 1 to 5 and 7 is not in the dandiset. All files

Reproducing Semenov et al. (2026) from dandiset 001521
reproduce_semenov_2026.ipynbExecuted, with outputs · Source

Verification

Round 1
2026-11-02
Agreed with 26 of 27 outcomes. Independent agent run by CatalystNeuro (same harness, separate context, did not write the notebook). Reviewed: Round 1. The executed notebook with the conclusion cells and the summary removed, the helper module and the claims file, with the paper. The verifier queried the archive API, rendered Fig. 6 of the paper, and tested the helper functions on synthetic stacks. Report in transcripts/001521/verifier-1-report.md. Its other points were also addressed: the docstring file counts are corrected (31 and 42), the conversion is now taken from Fig. 6B, and results with fields of view as the unit are added.
- fig6b-650-weak-but-significant: the verifier’s outcome was “Not reproduced”. Both tests the statement names are fully specified, and both conversions fail at least one. Under M1 the count matches under one rule, but the end mean is -9.2%, significant in the wrong direction, and the single-tailed test against 230 V/cm gives p = 0.20; under M2 against zero p = 0.27.
Round 2
2026-11-03Agreed with 27 of 27 outcomes. Independent agent run by CatalystNeuro (same harness, separate context, did not write the notebook). Reviewed: Round 2. The full executed notebook with the conclusions and summary, the README and this record. Report in transcripts/001521/verifier-2-report.md. In reply to its section 2, the notebook and record now print and cite the p-values that were given as below a bound (4.9e-09 and 3.0e-11 against the sham), the README states the 12 to 16% decline as the background-corrected dF and the raw decline as 7.5 to 8.7%, the summary and data check say the decline is steepest in the first recording and has no steady trend later in the train session, the train result is qualified to the 10-frame window with the 5-frame case (27.9%, higher fields 76.6 and 82.1%) given, the guess about Fig. 5B is removed, the file count is corrected to 33 trains and 9 shams in the train session, and outputs were added for the subject age and species and for the six other dandisets that name the paper (archive metadata). In reply to its section 3, a background taken as the 10th percentile of the cell-free pixels, less affected by dim unsegmented cells, was added; it changes no outcome. No outcome changed and none is proposed.

Method differences

How the tests differed from the paper’s method or data, and the claims each applies to.

The files give generator voltages and the paper field strengths, with no conversion in either. Two readings are run. M1, 3.9 V/cm per V, is adopted as the main reading (an assumption, as the paper gives no conversion), because the single-pulse voltages then fall on the paper's field series 570 to 1400 V/cm and the six train voltages nearest the fields labelled in Fig. 6B (230, 570, 650, 780, 900, 1000 V/cm) have group sizes close to the n the figure gives (28, 38, 36, 68, 47, 39 against 28, 37, 34, 54, 35, 36). M2, 3.3 V/cm per V, puts the train voltages 70, 200 and 230 V on the 230, 650 and 780 V/cm named in Section 3.6 and is shown as a variant. A named field is assigned to the nearest voltage. Neither factor follows from the electrode gap (V/d is 10 V/cm per V for the 1.00 mm of the files, 6.25 for the 1.6 mm of Methods 2.2). The outcome is the same under both.fig6a-ca-entry-1000-1200 fig6a-ca-sustained fig6a-780-not-significant fig6b-train-threshold fig6b-650-weak-but-significant
Cells are found by an Otsu threshold and watershed of the pre-pulse mean image, with a circular region of 0.7 of each cell's equivalent radius and the median of pixels more than 15 px from any cell as background; the paper drew the regions by hand (Methods 2.4) and these are standard choices, shown with regions of 0.5 and 1.0 of the radius, end windows of 5 and 20 frames, and a background taken as the 10th percentile of the cell-free pixels.fig6a-ca-entry-1000-1200 fig6a-ca-sustained fig6a-780-not-significant fig6b-train-threshold fig6b-650-weak-but-significant
Every detected cell in every file is used, as the dataset does not say which cells the paper used.fig6a-ca-entry-1000-1200 fig6a-ca-sustained fig6a-780-not-significant
The paper names no test for this panel; the claims file's readings (one-sample t-test or Student's t-test against 0 V/cm, end value or peak) are all run.fig6a-ca-entry-1000-1200
The cells were tested more than once in the dataset; all exposures are used, and the first exposure of each field of view is shown as a variant.fig6b-train-threshold
Reading (b), above 3 s.d. of the reference group, is run as above its mean plus 3 s.d. and, literally, above 3 s.d.fig6b-train-threshold

Provenance

Attempt record: 001521-semenov-2026-01
Paper: Sub-microsecond optical measurements of cell membrane charging and lesioning by pulsed electric fields
Date: 2026-11-02
Made by: An agent, run by CatalystNeuro
Dataset: DANDI:001521
Version: 0.261006.1845 (doi:10.48324/dandi.001521/0.261006.1845)
Run on: GitHub Actions
Harness: Claude Code
Model: claude-opus-5-5
Protocol: pilot-13

Transcripts of the agent runs

Each run’s prompt, messages and tool calls, with long output cut and the text of the paper left out, and its final report.

Other datasets that name the same paper

← DANDI:001518 DANDI:001525 →