Claims 41
From the paper Areal specializations in the morpho-electric and transcriptomic properties of primate layer 5 extratelencephalic projection neurons · claims file · checked 2026-10-22: ready
- cohort-animals
Tissue came from 58 macaques, 23 male and 35 female, 48 Macaca nemestrina and 10 Macaca mulatta, aged 2 to 19 years.
Male ( n = 23) and female ( n = 35) Macaca Nemestrina ( n = 48) and Macaca Mulatta ( n = 10) from 2 to 19 years old
STAR Methods, Experimental model and study participant details, Macaque specimensHow it will be judged
Counted over the subjects of the dataset's files, by subject identifier, sex and species. The 58 animals include donors whose tissue gave no recording in this dataset (the Methods count animals used in the study); the count of animals with at least one file is reported against 58.
- cohort-recordings
The physiology analysis rests on 333 recordings, 300 of them new in this study, plus 20 cells from a previous M1 study and 13 from a Patch-seq methods study.
For the current study, we presented data from 300 new patch-clamp/Patch-seq experiments, 168 of which were included in transcriptomic analyses. We also included data from 20 cells from a previously published study on M1 ... and 13 cells from a Patch-seq methods manuscript
Results, Patch-seq reveals subclass and areal distinctions, paragraph 2; STAR Methods, Neurophysiology features, paragraph 1How it will be judged
Counted as cells with current clamp data in the dataset. Two readings of what the dataset should hold, both reported, are (a) all 333 cells; (b) the 300 new cells only, as the 33 earlier cells were published with other datasets. Reproduced when the count matches either reading within tolerance.
- patchseq-qc-177
177 Patch-seq samples passed transcriptomic quality control (66 failed) and mapped to infragranular glutamatergic types of the reference taxonomy.
Using this approach, we collected 177 high-quality Patch-seq samples
Results, Patch-seq reveals subclass and areal distinctions, paragraph 1; Figure 2D, 2E; Figure S2AHow it will be judged
QC as in STAR Methods, Patch-seq sample mapping. A sample passes when its normalized marker sum (NMS) exceeds 0.7 and its correlation score to the MTG reference is at least 0.525, after correlation-based mapping with scrattch.mapping to the macaque MTG reference (Jorstad et al. 2023).
- cs-map-to-l5et
Retrogradely labelled corticospinal neurons map to the L5 ET subclass, 15 of 16 with the MTG reference and 16 of 16 with the M1 reference.
Retrogradely labeled CS neurons mapped to the L5 ET subclass in both the MTG (15 of 16 samples) and M1 (16 of 16 samples) reference taxonomy
Results, Patch-seq reveals subclass and areal distinctions, paragraph 1; Figure 2F; Figure S2EHow it will be judged
The CS samples are those tdTomato-positive after the spinal AAV-retro injection that passed QC (cell_counts.ipynb selects projection_type CS among QC-passing samples). Reproduced when the counts match within tolerance.
- fig3a-cs-with-l5et
In the UMAP of the physiology features, CS neurons lie only among the transcriptomic L5 ET neurons and overlap little with non-ET neurons, regardless of preparation, area and species.
CS neurons (triangles) exclusively distributed with the transcriptomically defined L5 ET neurons ( Figure 3A ) and had little overlap with non-L5 ET neurons. This separation did not depend upon slice preparation type
Results, Patch-seq reveals subclass and areal distinctions, paragraph 3; Figure 3A to 3C; Figure S3AHow it will be judged
The embedding as in STAR Methods, Physiology UMAP and classification, and UMAPprojections_020724.ipynb: the 40 features of the code's list; cells with more than 75% of features missing excluded; missing values imputed as the mean of the 2 nearest neighbours (KNNImputer) and features z-scored, separately for acute and cultured cells; PCA, keeping components that explain at least 1% of variance; UMAP with 25 neighbours, min_dist 0.1, Euclidean metric, random_state 20. The paper gives no measure of "exclusively distributed with". Two readings, both the reader's, both run: (a) in the two-dimensional embedding, the nearest transcriptomically labelled non-CS neighbour of every CS neuron is an L5 ET neuron; (b) in the space of the retained PCs, at least 9 of the 10 nearest labelled non-CS neighbours of each CS neuron are L5 ET, for at least 90% of CS neurons. Reproduced when both hold; partially when one does. The same rule is reported within each preparation, area and species.
- fig3e-classifier-accuracy
A logistic regression classifier trained on physiology predicts transcriptomic subclass (L5 ET, L5 IT, L5/6 NP, L6 IT) with 92.8% accuracy weighted across subclasses on the whole dataset, 87.9% holding out 50% and 82.2% holding out 75%; for L5 ET it is 98.6% and 93.8% (75% held out).
Classifiers trained and tested on 100% of the dataset had an average accuracy of 92.8% weighted across subclasses ... Classification was especially robust for the L5 ET subclass, where accuracy was 98.6% when trained on 100% of the dataset and dropped to 93.8% when holding out 75% of the dataset for cross-validation.
Results, Patch-seq reveals subclass and areal distinctions, paragraph 4; Figure 3E to 3GHow it will be judged
Training cells are QC-passing Patch-seq samples and CS neurons (labelled L5 ET); subclasses with 5 or fewer cells are dropped (classifier_features_070523.ipynb), leaving L5 ET, L5 IT, L5/6 NP and L6 IT. Features preprocessed as in fig3a-cs-with-l5et. Multiclass logistic regression on the feature z-scores; 100 repeats each of training on 100%, a random 50% and a random 25% of cells, testing on the held-out cells (on all cells for 100%), predictions aggregated. Open, as predict_subclass was not deposited: the regularization and solver (standard choice: scikit-learn defaults, L2 with C = 1, lbfgs, max_iter raised until convergence) and whether the random split is stratified (both run). The score: the Results say "average accuracy weighted across subclasses", the Methods "average f1 scores", and the code prints per-class f1 (0.986 for L5 ET at 100%, matching the 98.6%). Readings, all run: (a) support-weighted mean of per-class f1; (b) per-class recall (diagonal of the row-normalized confusion matrix) averaged with class weights. Reproduced when, under one reading, every value is within tolerance.
- fig3-spc-classifier
Classifiers built on sparse principal components of the physiology features perform similarly to those built on the features.
Similar results were obtained by building classifiers using sparse principal components (sPC) of physiology features
Results, Patch-seq reveals subclass and areal distinctions, paragraph 4; Figure S3BHow it will be judged
Sparse PCA with 8 components, alpha 7, random_state 0 on the preprocessed features (classifier_features_070523.ipynb), then the procedure of fig3e-classifier-accuracy on the 8 component scores. "Similar" has no measure in the paper; the reader's reading is that each of the five scores of fig3e is within 10 percentage points of the feature-based score computed here. Reproduced when all five are; partially when three or four are.
- fig3h-projection-largest-effect
In three-way ANOVAs of each physiological feature on projection type, preparation and cortical area, effect sizes are on average largest for projection type.
While preparation type and cortical area contributed to some variance, effect sizes, on average, were largest for neuronal projection type
Results, Patch-seq reveals subclass and areal distinctions, paragraph 5; Figure 3HHow it will be judged
Cells with a subclass (transcriptomic, CS, or predicted by the classifier trained on all labelled cells; predicted_subclass in L5_physiology_121223.csv), projection class L5 ET against all others. For each of the 40 features, OLS with projection class, preparation, area and all their interactions, type II sums of squares (UMAPprojections_020724.ipynb); Figure 3H plots the median effect size per factor over the 40 features. Effect size as in the Methods, partial eta squared SS_effect/(SS_effect + SS_residual); the code's formula (opening comment) is run as a variant. Reproduced when the median partial eta squared over features is larger for projection class than for preparation and for area.
- fig4a-top-subthreshold
Among features differing between L5 ET and non-ET neurons, the largest effect sizes are for subthreshold, HCN-related properties, resonant frequency, input resistance and the 3 dB cutoff.
The features with the largest effect size were those related to subthreshold membrane properties and HCN conductance (resonant frequency, input resistance, and 3dB cutoff).
Results, L5 ET and non-ET neurons are differentiated, paragraph 1; Figure 4AHow it will be judged
From the ANOVAs of fig3h-projection-largest-effect, the features whose projection effect has BH-corrected p at most 0.05 (the code corrects across the 40 features for each main effect) ranked by the projection effect size. "Input resistance" can be the maximum or the steady-state input resistance (both are features); either counts. Reproduced when the three highest-ranked features are resonant frequency, an input resistance and the 3 dB cutoff; partially when two of them are in the top three and the third in the top five.
- fig4c-et-higher-resonance
L5 ET neurons have a higher resonant frequency than non-ET neurons in both the MCx and the TCx, and in both acute and cultured slices.
L5 ET neurons had a higher resonant frequency ... and a lower input resistance than non-L5 ET neurons regardless of other factors
Results, L5 ET and non-ET neurons are differentiated, paragraph 1; Figure 4B, 4C; Figure S3CHow it will be judged
Resonant frequency as defined in the opening comment, on the cells of Figure 4C (147 L5 ET and 177 non-ET with the feature). Reproduced when the projection effect in the three-way ANOVA of fig3h has BH-corrected p < 0.05 and the L5 ET mean exceeds the non-ET mean within each area and within each preparation; partially when the effect holds but the direction fails in one area or preparation. Cell counts are reported against 147 and 177.
- fig4e-et-lower-rin
Maximum input resistance is lower in L5 ET than in non-ET neurons in both the MCx and the TCx, and in both preparations.
Maximum input resistance is lower in L5 ET neurons ( n = 145) than non-ET neurons ( n = 173) in both the MCx and TCx.
Figure 4D, 4E legend; Figure S3CHow it will be judged
Maximum input resistance as defined in the opening comment, on 145 L5 ET and 173 non-ET cells. Reproduced when the projection effect of the three-way ANOVA has BH-corrected p < 0.05 and the L5 ET mean is below the non-ET mean within each area and each preparation; partially when the effect holds but the direction fails in one group.
- fig4g-et-faster-spikes
L5 ET neurons have faster action potential kinetics than non-ET neurons, with larger maximum and minimum dV/dt and a shorter half-width, and faster repolarization in both areas and preparations.
exhibiting faster action potential kinetics ... Action potential repolarization was faster in L5 ET neurons compared with non-ET neurons in both neocortical areas and preparations
Results, L5 ET and non-ET neurons are differentiated, paragraph 1; Figure 4F, 4G; Figure S3CHow it will be judged
Max dV/dt, min dV/dt and half-width of the first spike at rheobase of the 1 s steps, as defined in the opening comment; Figure 4G has 148 L5 ET and 176 non-ET cells. For each of the three features the projection effect of the three-way ANOVA (BH-corrected p < 0.05) with L5 ET faster (larger max dV/dt, more negative min dV/dt, shorter half-width). Reproduced when all three hold and min dV/dt is faster in L5 ET within each area and each preparation; partially when min dV/dt holds and one of the other two does not.
- fig4-interaction-small
Projection type by area interactions are small; no feature differs strongly between L5 ET and non-ET neurons in one area and not the other.
Notably, projection type-by-area interactions were small, suggesting that no features were strongly different between L5 ET and non-ET neurons in one area and not the other
Results, L5 ET and non-ET neurons are differentiated, paragraph 1; Figure S3DHow it will be judged
A claim stated as small effect sizes, not as a test. From the three-way ANOVAs of fig3h, the projection by area interaction effect size of each feature. The paper gives no threshold; the reader's readings, both run, are (a) every interaction partial eta squared is below the median projection main effect over the 40 features; (b) for every feature whose interaction has p < 0.05 (uncorrected, as the code selects them for Figure S3D), L5 ET and non-ET differ in the same direction in both areas. Reproduced when both hold.
- species-no-effect
A four-way ANOVA on species, preparation, area and subclass finds no significant effect of species on any of the 40 physiological features.
Initial analyses involving a 4-way ANOVA with species, preparation, area, and subclass as the factors revealed no significant effects of species on any of the 40 physiological features extracted for analysis.
STAR Methods, Neurophysiology features, last paragraphHow it will be judged
A claim of absence on the paper's test, a four-way ANOVA per feature. The code for it was not deposited. Taken from the code's other ANOVAs: OLS with all interactions, type II sums of squares, BH correction across the 40 features for the species main effect (the Methods say p values were FDR adjusted). "Subclass" is open: (a) projection class, L5 ET against non-ET, as in the three-way ANOVA; (b) the predicted subclass. Species from the subject metadata. Reproduced when the corrected species p is at least 0.05 for every feature under the reading run; when the readings differ the claim is blocked as methods_underspecified. The smallest species p and its feature are reported, and the uncorrected count is shown as a variant.
- tcx-et-types-no-difference
Within TCx, physiological features do not differ significantly between L5 ET transcriptomic types (FDR-corrected p > 0.066), though some effect sizes are large.
Although we did not detect statistically significant differences between L5 ET types (FDR-corrected p > 0.066), effect sizes for some features were large
Results, L5 ET and non-ET neurons are differentiated, paragraph 2; Figure S3G, S3HHow it will be judged
A claim of absence. TCx L5 ET neurons with a t-type from the MTG mapping (cluster_Corr_label_MTG in cell_list_L5.csv), the types with enough cells (Figure S3H names ET_1 with 25 and ET_2 with 19). The test is not named; the Methods list ANOVAs, t-tests, Mann-Whitney U tests and BH correction. Readings, all run: (a) two-sided Welch t-test per feature, as the code's other two-group comparisons, BH across the 40 features; (b) two-way ANOVA on type and preparation, type effect, BH across features. Reproduced when no feature reaches corrected p < 0.05 under every reading; not reproduced when one does under every reading; blocked as methods_underspecified when the readings differ. The smallest corrected p and the largest Cohen's d are reported.
- fig4h-patchseq-genes-et
In the Patch-seq data, 47 ion channel, neurofilament and axonal targeting genes differ between L5 ET and non-ET neurons (partial eta squared at least 0.05), 21 of them also found in the snRNA-seq data, including HCN1 and PEX5L higher in L5 ET.
we identified 47 ion channel, neurofilament, and axonal targeting genes from our Patch-seq data with significantly different expression between L5 ET and non-ET neurons
Results, L5 ET and non-ET neurons are differentiated, paragraph 3; Figure 4H; Figure S4A to S4CHow it will be judged
As in STAR Methods, Gene expression analysis, and Interareal_Project_Patch_seq_Tdata_Comparisons_wStats.ipynb. log2(CPM + 1) of the 442 curated genes (gene_list.csv); genes with at least 0.5 log2 difference of medians between L5 ET and non-ET; three-way ANOVA on area, projection type and preparation; projection effect FDR-corrected p < 0.05 and effect size at least 0.05 (standard partial eta squared, the code's formula as variant). Reproduced when the counts are within tolerance.
- patchseq-genes-area
In the Patch-seq data, 43 curated genes differ between L5 ET neurons of MCx and TCx (partial eta squared at least 0.05), 10 of them also found in the snRNA-seq data, among them KCNN2, KCNN3, KCNC2, DPP6 and CACNA1A.
We identified 43 genes with significantly different expression and with η 2 ≥ 0.05 between L5 ET neurons from the MCx versus TCx
Results, Regional morpho-electric specializations in the L5 ET subclass, paragraphs 1 and 5; Figure S4D to S4F; Figure S6EHow it will be judged
As fig4h-patchseq-genes-et, restricted to L5 ET samples, with the 0.5 log2 difference between areas and a two-way ANOVA on area and preparation (Figure S4E), area effect, BH-corrected p < 0.05 and effect size at least 0.05. The overlap is counted against the genes of fig1-snrna-genes-area. Reproduced when both counts are within tolerance.
- fig1-snrna-genes-et
In the human and macaque snRNA-seq references, 29 curated genes differ between L5 ET and non-ET subclasses (at least 0.5 log2-fold, partial eta squared at least 0.05), including KCNN1, KCNN2, SCN4B, HCN1 and NEFH.
we identified 29 genes with ... between L5 ET and non-ET neuronal subclasses
Results, Differences in ion channel- and neurofilament-related gene expression, paragraph 3; Figure 1D; Figure S1E, S1FHow it will be judged
As in STAR Methods, Gene expression analysis, and RNA_seq10x_human and macaque.ipynb. 50 nuclei sampled per subclass, species and area (random_state 1); log2(CPM + 1); at least 0.5 log2 difference of medians; three-way ANOVA (area, projection type, species), type II; projection effect BH-corrected p < 0.05 and effect size at least 0.05 (standard partial eta squared, the code's formula as variant). Reproduced when the count is within tolerance.
- fig1-snrna-genes-area
Within the L5 ET subclass, 30 curated genes differ between M1 and MTG in the snRNA-seq references, including SK channel genes, KCNC2 and KCNIP1.
We identified 30 genes with significantly different expression and with ... between L5 ET neurons in M1 versus MTG datasets
Results, Differences in ion channel- and neurofilament-related gene expression, paragraph 4; Figure 1E; Figure S1GHow it will be judged
As fig1-snrna-genes-et, restricted to L5 ET nuclei, with the 0.5 log2 difference between areas and a two-way ANOVA on area and species, area effect.
- fig5-tap-root-dendrites
Several MCx L5 ET neurons, including CS neurons, have long "tap root" basal dendrites reaching about 1.4 mm from the soma; TCx L5 ET neurons have none but have more complex apical tufts.
possessed prominent ... basal dendrites that, in the most extreme cases, extended ~1.4 mm from the soma
Results, Regional morpho-electric specializations in the L5 ET subclass, paragraph 2; Figure 5A; Figure S5A, S5BHow it will be judged
On the 12 MCx and 11 TCx reconstructions. The paper defines no measure of a tap root, and "extended ~1.4 mm from the soma" can be read as (a) the maximum Euclidean distance of a basal dendrite tip from the soma or (b) the maximum basal path distance (basal_dendrite_max_path_distance, a feature of the morphology code); both are run. Reproduced when, under one reading, the MCx maximum is within tolerance of 1.4 mm and every TCx value is below the MCx median (the reader's measure of "lacked tap root dendrites altogether"; the paper gives none).
- fig5b-dendrite-depth
L5 ET neurons in TCx have more dendrite in L1, those in MCx more dendrite in deeper layers.
The total length of dendrites in each layer differed between neocortical areas, with L5 ET neurons in the TCx having more dendrites in L1 and L5 ET neurons in the MCx having more dendrites in deeper layers
Results, Regional morpho-electric specializations in the L5 ET subclass, paragraph 2; Figure 5BHow it will be judged
Dendrite length per layer from the reconstructions aligned to the pia-white matter axis (skeleton_keys). Layer boundaries are not given for these slices; the paper's own layer-aligned features are the reading. The paper reports no test; reproduced when the mean L1 dendrite length is larger in TCx and the mean length in L5 and L6 is larger in MCx.
- fig5c-basal-apical-features
MCx L5 ET neurons have longer basal dendrites extending farther horizontally, and TCx L5 ET neurons have more apical branches and a higher maximum apical branch order.
Basal dendrites were longer (FDR-corrected p = 0.0008, two-sided t test, d = 2.22) and extended farther in the horizontal direction (FDR-corrected p = 0.0006, two-sided t test, d = 2.65) in the MCx compared to the TCx, and L5 ET neurons in the TCx had both more apical dendrite branches (FDR-corrected p = 0.009, two-sided t test, d = 1.56) and higher-order apical dendrite branches
Results, Regional morpho-electric specializations in the L5 ET subclass, paragraph 2; Figure 5C, 5DHow it will be judged
Features from skeleton_keys and neuron_morphology after shrinkage correction (STAR Methods, Morphology feature analysis). As in morphology.ipynb, Welch t-test per feature, BH correction across all features (apical depth PCs and axon features excluded), absolute Cohen's d with pooled SD. The features are read as basal_dendrite_max_path_distance (or total length, both run) for "longer", basal_dendrite_extent_x for "horizontal", apical_dendrite_num_branches and apical_dendrite_max_branch_order. The deposited notebook printed other values (extent_x corrected p 0.0015, d 2.79; max branch order d 2.11), so the paper's numbers came from another run. Reproduced when all four differ at corrected p < 0.05 in the stated direction; the d values are compared with tolerance.
- fig5e-morph-classifier
A classifier predicts the area of an L5 ET neuron from dendritic features with 100% accuracy on the whole dataset, 91.9% weighted f1 holding out 50% and 76.1% holding out 75%.
the classifier obtained an accuracy of 100% when trained and tested on the whole dataset and maintained better-than-chance levels of performance when holding out 50% (average weighted f1 score = 91.9%) or 75% (average weighted f1 score = 76.1%)
Results, Regional morpho-electric specializations in the L5 ET subclass, paragraph 3; Figure 5EHow it will be judged
As in morphology.ipynb, imputation and scaling as for physiology, logistic regression, 100 repeats per training fraction (100%, 50%, 25%), weighted f1. Classifier settings open as in fig3e-classifier-accuracy. The deposited notebook printed 0.900 and 0.741.
- fig6a-area-variation
Two-way ANOVAs (preparation and area) show substantial cross-areal variation in subthreshold and suprathreshold properties of L5 ET neurons.
This analysis revealed substantial cross-areal variation in subthreshold and suprathreshold properties
Results, Regional morpho-electric specializations in the L5 ET subclass, paragraph 4; Figure 6AHow it will be judged
L5 ET neurons as in the code (predicted_subclass L5 ET, L5ET_121223.csv; Figure 6 has 77 or 78 MCx and 70 TCx cells). For each of the 40 features, OLS on area, preparation and their interaction, type II, BH correction across features per term (L5ET.ipynb; its feature list now also holds three AHP features, AHP_area, AHP_duration and AHP, which the deposited ET_cross_p.csv does not hold, so the 40 are the reading). The code's corrected area p values (ET_cross_p.csv) are below 0.05 for 19 of the 40 features, among them Fr, cutoff, zap_tau, initial_ifreq_gain, max_ifreq_gain, up_down_ratio, f_AHP, MaxRn, SSRn, MaxZ, AP_HW, min_dvdt and threshold_current. "Substantial" has no measure; reproduced when at least one subthreshold (chirp or input resistance) and one suprathreshold (spike or firing) feature have corrected area p < 0.05, and the count of such features is reported against 19.
- fig6b-mcx-higher-resonance
L5 ET neurons in the MCx have a higher resonant frequency and a higher 3 dB cutoff than those in the TCx.
Resonant frequency and the 3dB cutoff in response to the chirp stimulus were higher in L5 ET neurons in the MCx compared to the TCx
Results, Regional morpho-electric specializations in the L5 ET subclass, paragraph 4; Figure 6BHow it will be judged
From the two-way ANOVAs of fig6a, the area effects for resonant frequency (77 MCx, 70 TCx in Figure 6B) and 3 dB cutoff. Reproduced when both have BH-corrected p < 0.05 with the MCx mean higher; partially when one does. The paper's code obtained means of 5.16 and 3.30 Hz (fR) and 10.99 and 8.10 Hz (cutoff), MCx and TCx, corrected p 5.7e-5 and 3.1e-4 (ETmean_sem.csv, ET_cross_p.csv); these are reported alongside, not as targets.
- fig6d-mcx-deeper-fahp-symmetric
L5 ET neurons in the MCx have a more pronounced fast AHP and faster, more symmetrical action potentials (lower ratio of max to min dV/dt) than those in the TCx.
L5 ET neurons in the MCx had a more pronounced fast afterhyperpolarization (fAHP) and exhibited faster, more symmetrical ... action potentials compared to those in the TCx
Results, Regional morpho-electric specializations in the L5 ET subclass, paragraph 4; Figure 6C to 6EHow it will be judged
The Figure 6D and 6E legend says the opposite for the fAHP and for symmetry ("TCx ... have a deeper fast AHP ... than ... MCx ... and are more symmetrical"); the Results and the code's means (fAHP -14.18 mV MCx, -10.74 mV TCx; up/down ratio 2.84 and 3.57) agree with each other, so the Results are tested and the legend is recorded as a paper inconsistency. From the ANOVAs of fig6a: fAHP (more negative relative to threshold in MCx), up/down ratio (lower in MCx), and "faster" read as (a) shorter half-width and (b) more negative min dV/dt (the code finds max dV/dt not different, corrected p 0.13). Reproduced when fAHP, the up/down ratio and one reading of "faster" have corrected area p < 0.05 in the stated direction; partially when two of the three do.
- figs6b-mcx-lower-rin
The input resistance of L5 ET neurons is lower in MCx than in TCx, in both acute and cultured slices.
the input resistance of L5 ET neurons is lower in MCx (n = 76) compared with TCx (n = 68) in both acute and culture preparations (FDR corrected p = 0.003, area effect, 2-way ANOVA)
Figure S6B legend, leftHow it will be judged
The legend does not say which input resistance; the code's corrected area p is 0.0033 for steady-state (SSRn) and 0.0081 for maximum input resistance (MaxRn), so SSRn matches the printed p. Both are run. Reproduced when the area effect has corrected p < 0.05 with MCx lower, and MCx is lower within each preparation, for SSRn; MaxRn is shown as a variant.
- figs6b-mcx-higher-rheobase
MCx L5 ET neurons need more current to start spiking than TCx L5 ET neurons.
the current required to generate spiking in L5 ET neurons is higher in MCx (n = 78) compared to TCx (n = 70; FDR corrected p = 0.0496, area effect 2-way ANOVA)
Figure S6B legend, middle; Results, Regional morpho-electric specializations, paragraph 5How it will be judged
Rheobase as the lowest 1 s step amplitude that evokes a spike (the code's threshold_current; its corrected area p is 0.049). Reproduced when the area effect has corrected p < 0.05 with MCx higher. As p is near 0.05, the uncorrected p and the effect size are reported.
- figs6b-avg-gain-no-difference
The slope of the average-rate FI curve (maximum gain) does not differ between MCx and TCx L5 ET neurons.
the slope of the average firing rate FI curve (max gain) is not different between areas (MCx, n = 78; TCx, n = 70; FDR corrected p = 0.869, area effect 2-way ANOVA)
Figure S6B legend, right; Results, Regional morpho-electric specializations, paragraph 5How it will be judged
A claim of absence on the paper's test, the area effect of the two-way ANOVA with BH correction across features. Maximum gain as in the code, the largest step-to-step increase in spike count of the 1 s steps divided by the current step (Hz/pA). The code's corrected p is 0.850, not the printed 0.869. Reproduced when corrected p is at least 0.05; the means and partial eta squared are reported either way.
- fig6h-tcx-steeper-instantaneous-fi
TCx L5 ET neurons fire onset doublets or bursts, so both the initial and the maximum slope of the maximum instantaneous FI curve are higher in TCx than in MCx.
In the TCx, L5 ET neurons exhibited doublets or high-frequency bursts of action potentials at the onset of the current that were rarely observed elsewhere in the sweep or in MCx L5 ET neurons
Results, Regional morpho-electric specializations in the L5 ET subclass, paragraph 5; Figure 6F to 6HHow it will be judged
The paper quantifies the bursts through the slopes of the maximum instantaneous rate (1/ISI min) against current on the 1 s steps: initial gain is the slope where the neuron begins firing (the first suprathreshold step and the step before it) and maximum gain the largest step-to-step slope, each divided by the current step (Figure 6H has 69 TCx and 78 MCx). From the ANOVAs of fig6a, reproduced when both have corrected area p < 0.05 with TCx higher; partially when only the initial slope does. The code obtained corrected p 5.7e-5 and 4.7e-4 and initial-slope means of 0.28 (TCx) and 0.10 (MCx) Hz/pA.
- figs6c-mcx-higher-max-rate
At maximal current injections MCx L5 ET neurons sustain higher average firing rates than TCx L5 ET neurons.
L5 ET neurons in the MCx sustained higher average spike counts compared to L5 ET neurons in the TCx at maximal current injections ... maximal average firing rate, p = 0.0002, two-sided t test, d = 1.0
Results, Functional specializations of MCx L5 ET neurons, paragraph 2; Figure S6CHow it will be judged
Cells: the L5 ET neurons recorded with 1 s steps increasing by 200 pA up to 4 nA (STAR Methods), 42 MCx and 26 TCx (keys of MCx.pkl and TCx.pkl). Per cell the maximum over steps of the spike count in the 1 s step; Welch t-test and pooled-SD Cohen's d (ExtendedFI.ipynb, which printed means 101.6 MCx and 76.0 TCx, p 0.00024, d 1.00). Reproduced when p < 0.05 with MCx higher and d within tolerance; partially when only the test holds.
- figs6d-equal-max-instantaneous
TCx L5 ET neurons reach higher instantaneous rates below 1 nA, but at up to 4 nA MCx L5 ET neurons fire at equivalent maximal instantaneous rates.
MCx neurons were capable of firing equivalently at higher current injection amplitudes ... maximal instantaneous firing rate, p = 0.99, two-sided t test, d = 0.003
Results, Functional specializations of MCx L5 ET neurons, paragraph 2; Figure S6DHow it will be judged
A claim of absence on the paper's test. Per cell the maximum over all steps of the maximum instantaneous rate 1/ISI(min); Welch t-test (ExtendedFI.ipynb printed 243.8 Hz TCx and 244.1 Hz MCx, p 0.99). Reproduced when p is at least 0.05; Cohen's d is reported. The first clause (TCx higher below 1 nA) rests on Figure S6D alone with no test; the mean maximum instantaneous rate per area at each step below 1 nA is shown and does not decide the outcome.
- fig6-depolarization-block
At high current amplitudes, TCx L5 ET neurons undergo depolarization block far more often than MCx L5 ET neurons (14 of 26 against 1 of 42).
L5 ET neurons in the TCx were more likely to undergo depolarization-induced block of action potential generation in response to these higher current injection amplitudes (MCx = 1 of 42 neurons; TCx = 14 of 26 neurons)
Results, Functional specializations of MCx L5 ET neurons, paragraph 2; Figure 6J, 6K and S6C, S6D (arrows)How it will be judged
Block is "a negative slope in the average rate FI curves" (STAR Methods). The readings are (a) any decrease of spike count from one step to the next; (b) the spike count at the largest step below the maximum of the curve. Reproduced when, under one reading, both counts are within tolerance (1 of 42 exactly, 13 to 15 of 26); partially when, under one reading, the TCx proportion exceeds the MCx proportion but a count is outside tolerance. The paper reports no test; a Fisher exact test is shown as a variant and does not decide the outcome.
- fig6j-spike-width-ratio
At the maximal average firing rate, spike width more than doubles from first to last spike in TCx L5 ET neurons but increases only 1.5 times in MCx L5 ET neurons.
At the maximal average firing rates, the spike width in TCx L5 ET neurons more than doubled, whereas MCx L5 ET neuron spike widths only increased 1.5 times ... p = 1.7e−09, two-sided t test, d = 2.02
Results, Functional specializations of MCx L5 ET neurons, paragraph 2; Figure 6JHow it will be judged
Per cell, the step with the most spikes; at that step the ratio of the half-width of the last spike to that of the first; Welch t-test and Cohen's d (ExtendedFI.ipynb printed 2.27 TCx, 1.53 MCx). The code takes the argmax over steps with nonzero spike counts and uses that index into the full array; when steps without spikes precede firing this picks a later step. Readings, both run: (a) the step with the most spikes; (b) the code's indexing. Reproduced when the TCx mean exceeds 2, the MCx mean is within tolerance of 1.5, p < 0.05 and d within tolerance, under one reading.
- fig6k-upstroke-ratio
At the maximal average firing rate, max dV/dt of the last spike falls to about 20% of the first in TCx L5 ET neurons but only to about 50% in MCx.
Similarly, the maximum dVdt decreased to ~20% of the first spike value in TCx neurons versus only ~50% in the MCx ... p = 3.2e−10, two-sided t test, d = 1.69
Results, Functional specializations of MCx L5 ET neurons, paragraph 2; Figure 6KHow it will be judged
As fig6j-spike-width-ratio, for the ratio of max dV/dt of the last to the first spike (ExtendedFI.ipynb printed 0.221 TCx and 0.520 MCx, p 3.2e-10, d 1.69). The values are "~20%" and "~50%"; reproduced when both means and d are within tolerance and p < 0.05 under one reading; partially when the test holds and one mean is outside tolerance.
- figs7-cv-isi-higher-mcx
During 10 s near-rheobase steps, MCx L5 ET neurons fire more irregularly (higher CV of ISIs) than TCx L5 ET neurons across a wide range of firing rates.
which revealed more variable ISIs in MCx L5 ET neurons across a wide range of average firing rates during the 10-s current step
Results, Functional specializations of MCx L5 ET neurons, paragraph 3; Figure S7F, S7GHow it will be judged
Sweeps of the 10 s steps with at least 10 spikes; spikes in the first 100 ms excluded; CV = SD/mean of the ISIs (STAR Methods); each sweep is an observation. The second factor of the two-way ANOVA is not named; Figure S7G bins CV by the number of spikes in the sweep, so the reading is area by spike-count bin. The bin edges are not given; readings, both run: (a) bins of 10 spikes; (b) bins of 25 spikes. The code was not deposited. Reproduced when the area effect has p < 0.001 with MCx higher under both readings; partially when p < 0.05; observation and cell counts are reported against the legend's.
- fig7-oscillation-frequency-vm
The peak frequency of subthreshold oscillations between spikes increases with membrane potential in MCx L5 ET neurons but not in TCx, by autocorrelation and by Fourier analysis.
However, the peak frequency of these oscillations increased with membrane potential only in the MCx, whether measured from autocorrelation
Results, Functional specializations of MCx L5 ET neurons, paragraph 3; Figure 7I, 7JHow it will be judged
Method from STAR Methods (the R code was not deposited): sweeps of the 10 s steps with at least one spike and at most 200; interspike segments from the return to baseline after the AHP (found by fitting a curve) to 50 ms before the next spike; filtered 3 to 50 Hz; segments longer than 650 ms analysed by FFT (peak frequency and strength per segment) and a smoothing spline over segments for the sweep's peak frequency and strength; autocorrelogram over the last 1 s of segments longer than 1 s, frequency from its peaks and troughs; segments aggregated into 5 mV bins of membrane potential per cell, each cell-bin an observation; Pearson correlation over observations. Open, with standard choices: the curve fitted to find the return to baseline (standard: an exponential fit to the AHP decay, baseline reached within 1 mV of its asymptote), the filter (standard: zero-phase 4th-order Butterworth band-pass), the spline (standard: R smooth.spline defaults), and the autocorrelogram frequency (standard: 1 over the lag of the first peak after zero lag). These could change the values; the readings are run as stated and the counts of observations reported against the legend's. Reproduced when the MCx correlations are positive with p < 0.001 and the TCx correlations have p at least 0.05, for both measures; the r2 values are compared with tolerance. TCx r2 below 0.001 with p 0.146 on 82 observations is not self-consistent (such an r2 gives p near 0.8), so only p is used for TCx.
- fig7k-oscillation-power-vm
Oscillation strength increases modestly with membrane potential, significantly in MCx L5 ET neurons and not significantly in TCx.
increased in power modestly as a function of membrane potential
Results, Functional specializations of MCx L5 ET neurons, paragraph 3; Figure 7KHow it will be judged
Method as fig7-oscillation-frequency-vm, with the spline-fitted FFT peak strength. Reproduced when the MCx correlation is positive with p < 0.001 and the TCx correlation is positive with p at least 0.05; partially when the MCx part holds only; the r2 values are compared with tolerance.
- figs7-oscillations-with-blockers
Subthreshold oscillations persist with fast synaptic transmission blocked, at a slightly reduced frequency, so they are intrinsic to the neuron.
Oscillations occurred in the presence of blockers of fast synaptic transmission, albeit at a slightly reduced frequency
Results, Functional specializations of MCx L5 ET neurons, paragraph 3; Figure S7D, S7EHow it will be judged
Blockers (DNQX with d-APV and gabazine or picrotoxin) per cell from the recording metadata. Oscillation frequency as in fig7-oscillation-frequency-vm. The legend bins by number of spikes in the sweep; the two-way ANOVA is read as blocker by spike-count bin, with the bin edges open (bins of 10 and of 25 spikes, both run). Reproduced when oscillations are detected in cells with blockers and the blocker effect has p < 0.05 with lower frequency under both readings; partially when oscillations are present but the effect is not significant.
- fig7l-firing-tracks-oscillation
In MCx L5 ET neurons the modal instantaneous firing rate during spiking epochs is well correlated with the subthreshold oscillation frequency; in TCx it is not.
Consequently, the instantaneous firing rate during spiking epochs was well correlated with the prominent subthreshold oscillatory frequency in L5 ET neurons in the MCx
Results, Functional specializations of MCx L5 ET neurons, paragraph 4; Figure 7LHow it will be judged
Modal firing rate is the most common value of 1/ISI over the 10 s step (STAR Methods); the binning of instantaneous rates to find a mode is not given (readings, both run, 1 Hz and 2 Hz bins). Observations as in Figure 7L, each a sweep with an FFT peak frequency. Reproduced when the MCx correlation is positive with p < 0.001 and r2 within tolerance and the TCx p is at least 0.05; partially when the MCx correlation holds with r2 outside tolerance.
- figs7g-modal-rates-long-isis
MCx L5 ET neurons reach modal firing rates near 30 Hz while still having ISIs longer than 650 ms, whereas no TCx neuron with such ISIs has a modal rate above about 8 Hz.
In the MCx, L5 ET modal firing rates were as high as ~30 Hz, accompanied with ISIs longer than 650 ms
Results, Functional specializations of MCx L5 ET neurons, paragraph 4; Figure S7GHow it will be judged
Over the 10 s sweeps that contain at least one ISI longer than 650 ms, the largest modal firing rate per area, modal rate as in fig7l. The TCx value is a bound in the paper ("no TCx neurons had modal firing rates greater than ~8 Hz"), so reproduced when, under one binning reading, the MCx maximum is within tolerance of 30 Hz and the TCx maximum is at most 8.8 Hz (8 Hz plus its tolerance).
Pipeline
- Paper
- done (1 of 1)
- Claims
- done
- Claims check
- done (ready)
- Attempt
- todo
- Questions
- none
- Verified
- none