Linear fits to the monotonically increasing segment yielded conduction velocities of 0.2 to 0.27 m/s
Results, Millisecond pulse-locking dynamics differentiate learning-induced plasticity, paragraph 2; fig. S4, B and C
What the paper reports
- latency against distance, 4 uA: OLS regression on binned means, slope = 0.005 ms/um, r2 = 0.97, p < 0.001
- latency against distance, 5 uA: OLS regression on binned means, slope = 0.004 ms/um, r2 = 0.95, p < 0.001
- latency against distance, 6 uA: OLS regression on binned means, slope = 0.004 ms/um, r2 = 0.91, p = 0.001
- jitter against distance, 4 uA: OLS regression on binned means, p = 0.095
- jitter against distance, 5 uA: OLS regression on binned means, p = 0.43
- jitter against distance, 6 uA: OLS regression on binned means, p = 0.53
With the paper's mapping, PL latency rises with distance at every current: binned means over 120 to 840 um give slopes 0.00591, 0.00445 and 0.00442 ms/um (p = 0.000384, 0.000314, 0.00079; r2 0.93, 0.94, 0.91) and velocities 0.169, 0.225 and 0.226 m/s; binned medians give p = 0.000147, 0.000356, 0.000486 and velocities 0.161, 0.188 and 0.199 m/s. At 4 uA both velocities are below 0.18 m/s, so no reading has all three inside 0.18 to 0.297. Jitter shows no slope (binned p = 0.148, 0.33, 0.377; unbinned 0.79, 0.819, 0.143). Check of the mapping: the stimulation contacts lie at 0 to 520 um depth, 12 of 17 between 150 and 420 um (Methods 200 to 400 um), and the fraction of modulated responses, pooled over mice, falls from 0.711 at 60 to 120 um to 0.390 beyond 960 um; the check is not clean, as the 0 to 60 um bin is the lowest (0.315) and per mouse the fall reverses in Mouse 5 (0.31 within 240 um, 0.48 beyond 960 um) and Mouse 6 (0.08 and 0.88, not in the fit), while Mouse 5 looks local under the electrode-table mapping (0.68 and 0.56). Without Mouse 5 the 4 uA velocity is 0.179 m/s from binned means and medians, still below 0.18; leaving out one mouse at a time it is 0.146 to 0.189 m/s, and 0.18 or more under both aggregations only without Mouse 1 (0.189 and 0.180). With the electrode table's rel_y (round 1, variant) the contacts lie at -156 to 1559 um, the modulated fraction does not fall with distance, and no latency slope is significant (p 0.294 to 0.962).
How the test differed from the paper’s method
- Distance follows the paper's code (processing/batch_process/postprocessing/longitudinal_data_utils.py, add_data_to_df and correct_df_unit_location; python/fig_s4/generate_figure.py; XieLuanLab/icms-activation-plasticity at d1922d1): the contact at (32 - stim_channel) x 60 um up the probe, the unit at unit_y_um, both converted to cortical depth with the animal's length of probe outside the brain and cos 30 degrees, responses at or above the surface dropped, Mouse 6 left out (the code has no length for it), PL responses with the z cap of 100. Responses at 120 to 840 um are binned in 120 um bins from 60 um, as the code does.
- Pulse-locking follows the paper's code (processing/batch_process/postprocessing/responses_v2/pulse_locked_response_metrics.py, called from responses_v3/stim_condition_response.py, XieLuanLab/icms-activation-plasticity at d1922d1): 500 draws of 20% of the pulses without replacement, two points inserted at 1.5 ms, and a null of 100 uniform placements of the spikes of one 20% draw divided by the pulses drawn. The claims file's common method says only 'the same spike count'; round 1 used the spikes of all pulses, which narrowed the null and called 80% of responses PL. Pulses are taken 10 ms apart from the train onset, as the files store train onsets only (every 70-pulse train lasts 700.00 ms).