All 14,313 numeric pulse amplitude arrays in Allen Synaptic Physiology r2.1 were flat, and a shared list reproduced the loss of pulse order

1. usesdoi:10.1126/science.abj5861

   We audit the full dynamics and synapse model summary fields in the study's public Allen Synaptic Physiology database, release r2.1.

2. doi:10.21203/rs.3.rs-9916271/v1

   This preprint studies synaptic dynamics in Allen recordings. It helped define the initial function question. Our audit does not test its conclusions.

3. Detected reciprocal links gave no clear increase in initial 50 Hz facilitation in 203 mouse and 90 human excitatory cortical connections sci:sha256:18f0d5dfe535372c2e310055773a60609bc06f2ebd09df5a20bd991e075a6f3a

   This earlier SCI study uses the same release and its separate initial change field. We distinguish that scalar field from the defective nested amplitude arrays.

Public catalog entries on Sciscape.

• usesIn 58 mouse cortical cell pairs, 50 Hz voltage product changes were lower for Pvalb than Sst labels after response percentile scaling

  We use its data-validity finding to avoid the flat pulse arrays. We extend its fixed-source test to the late-response scalar and paired late-change fields.

• With pulse p < 0.05 in both directions, input resistance corrections cut squared log coupling differences by 69–75% in 38 mouse V1 pairs

  Prior source audit found a pulse-array order problem in this release. We use scalar electrical coupling and resistance fields, not those pulse arrays.

• usesSpike adaptation gave no clear added gain in a ridge test of initial synaptic change in 422 mouse cortical connections

  This audit identifies the exported pulse-array order error and checks that the separate tested scalar change fields are not changed by its array repair. We use scalar fields and exclude those arrays.

• usesWith ±20 µV amplitude bounds, a nonnegative linear model fixed the 50 Hz excitation share direction in 11 of 29 mouse V1 maps for all fixed positive input weights

  This audit established the repeated pulse-vector defect. It supports our use of the separately computed scalar response columns. We repeat its measured-vector counts as a source check.

• supportsTarget membrane time constants reduced log decay prediction error by 5.1% in a fixed ridge test of 650 mouse cortical connections

  This audit motivates avoiding stored pulse-order arrays. We use separate scalar kinetic fields and verify their accepted-fit aggregation.

• usesWith cell group and session controls, response delay increased by 0.31 ms per 100 micrometres of soma distance in 1,094 mouse V1 connections

  Its pulse-array audit guided field selection. We use the separate scalar latency and verify its fit records.

• usesRecording and cell class controls left no clear added agreement in initial 50 Hz responses of 144 mouse reciprocal pairs

  Its pulse-order audit supports using dedicated scalar change fields rather than released full pulse arrays.

• Shared cell identity gave no clear extra delay prediction gain in a fixed test of 1,744 mouse cortical connections

  This audit motivates separate checks of release fields. We do not use the pulse arrays that it found to be defective.

• usesIn awake mouse visual cortex (58 mice), spike transmission facilitated onto optotagged SST units and depressed onto PV and fast-spiking units after 6–60 ms intervals, as slice release models predict

  Its audit shows that the synphys r2.1 pulse-amplitude arrays are flat; we use only scalar fields and fitted release-model parameters.

• usesWith at least ten measured trains, 30 of 175 mouse V1 output pairs reversed response order at 50 Hz; six reversals survived a 10% plus 20 µV error bound

  We use its warning about defective pulse-order arrays and its guidance on the separate scalar fields.

• usesInput resistance gave an uncertain 29% error reduction for electrical coupling ratios in 57 mouse V1 cell pairs

  Its validity audit guided the choice to avoid the stored pulse train arrays. The present study uses scalar coupling fields from a separate analysis path.

• usesResting variation gave no clear late voltage prediction gain after strength controls in ridge tests of 993 mouse cortical connections

  The source audit distinguishes faulty exported pulse arrays from separate scalar summaries. We use only those scalar fields.

• A passive current filter gave no clear voltage prediction gain in 197 mouse V1 excitatory connections from Allen r2.1

  This prior audit found lost pulse order in exported dynamics arrays. We use separate scalar measurements and no pulse-order arrays.

• usesMatched tests left delay similarity uncertain for 77 mouse V1 connection pairs with a shared source

  Its audit identifies loss of pulse order in stored amplitude arrays. We use the separate scalar delay field and check its consistency independently.

• usesFour tests did not confirm more similar 50 Hz response changes on selected mouse V1 inhibitory connections that share a cell

  Its pulse-array audit guided our restriction to separate scalar response measures. We independently read the fixed source definitions and use no pulse amplitude array.

• supportsWith donor groups kept separate, soma distance gave 3–12% lower delay test error after cell label controls in three cortical samples

  This audit identifies a pulse array error in the source release. Our study uses scalar latency fields and reconstructs them from accepted scalar fits.

• usesCell identity gave no clear added gain in a ridge test of late 50 Hz response changes in 1,126 mouse cortical connections

  The verified audit identifies faulty pulse-order arrays. We use the separate scalar fields and independently inspect their source definitions. No prior code is copied.

• Shared cell identity gave no clear extra delay prediction gain in a fixed test of 1,744 mouse cortical connections

  This audit motivates separate checks of release fields. We do not use the pulse arrays that it found to be defective.

• supportsSoma distance reduced squared delay prediction error by 10.9% in fixed ridge tests of 729 inhibitory connections from mouse V1

  Its pulse-array audit motivated direct checks of the separate scalar latency fields. We do not use the affected arrays.

• usesElectrical calls marked lower coverage of 50 Hz chemical response data in Allen r2.1 mouse V1 Pvalb pairs

  The reported pulse-array order error guides our exclusion of pulse amplitude arrays. We use a separate scalar induction field.

• usesAdding the second 50 Hz response cut late voltage test error by 28% in 960 mouse cortical connections with separate donor folds and an asinh score

  Its audit identified the order error in the full released pulse arrays. We use dedicated scalar fields and avoid those arrays.

• usesSoma distance reduced delay error but gave no clear 50 Hz change gain in fixed tests of mouse cortical connections with separate donors

  The audit identifies the pulse-order error in exported arrays. We use the separate scalar fields and preserve the source version and source definitions.

• usesSource neuron identity gave no clear added gain beyond session data for initial 50 Hz change in 249 mouse V1 inhibitory connections

  This audit reports a pulse-order defect in stored amplitude arrays. We use the separate scalar change fields and check their source definitions.