Claims 36
From the paper A line attractor encoding a persistent internal state requires neuropeptide signaling · claims file · checked 2026-10-22: ready
- fig1e-oxtr-avpr1a-coexpression
About 78% of Esr1+ VMHvl cells express Oxtr mRNA, and about 61% of these also express Avpr1a mRNA, in a single-cell RNA-seq atlas of VMHvl.
Overall, ~78% of Esr1 + cells in VMHvl expressed Oxtr mRNA, and ~61% of these cells also expressed Avpr1a mRNAs
Results, Most Esr1+ neurons co-express Oxytocin and Vasopressin receptors, paragraph 1; Figure 1EHow it will be judged
Uses the outside atlas of Kim et al. 2019 (doi:10.1016/j.cell.2019.09.020), analysed "as described previously" (STAR Methods, Transcriptomic analysis); the expression threshold is not given. Reproduced when both fractions are within tolerance.
- fig1f-peptides-activate-slices
Bath application of OXT plus AVP (400 nM each), or of either peptide alone, evokes strong calcium responses (dF/F about 500-600%) in VMHvl Esr1 neurons in acute slices.
Perfusing a mixture of 400nM (each) OXT plus AVP or individually administering these peptides elicited strong (ΔF/F ~500–600%) responses in VMHvl Esr1 cells
Results, Most Esr1+ neurons co-express Oxytocin and Vasopressin receptors, paragraph 2; Figure 1F, S1F-GHow it will be judged
The paper reports a range, 500 to 600%; the claim holds inside the range widened by 10% of each bound (450 to 660%), for the mean response of the 124 cells of Figure 1F. Which measure of the response (peak of the mean trace, or mean of per-cell peaks) is not given; either reading counts.
- fig2e-editing-reduces-slice-responses
In slices from mice with Cre-dependent Oxtr/Avpr1a gRNAs, the fraction of GCaMP8s-expressing cells responsive to OXT plus AVP is reduced from 42.6 +/- 3% (scrambled gRNA) to 13 +/- 3.3%.
The fraction of GCaMP8s-expressing cells responsive to mixed AVP and OXT peptides was reduced from 42.6± 3% in Scr gRNA controls to 13± 3.3% (~70% reduction) in Oxtr/Avpr1a gRNAs
Results, Altered VMHvl Esr1 neural activity during social interactions, paragraph 1; Figure 2EHow it will be judged
The criterion for a responsive cell is not given in the Methods. Reproduced when both fractions are within tolerance (the paper's stated spread) and the Mann-Whitney test across slices gives p < 0.01, the level printed in the Figure 2 legend.
- fig2g-bilateral-editing-reduces-aggression
Bilateral co-editing of Oxtr and Avpr1a in VMH (LV gRNAs plus Cas9 AAV) reduces attack towards male intruders (number, probability, total time and bout duration of attacks) and lengthens the latency to the first attack.
Experimental mice displayed a notable reduction in aggression towards male intruders, as evidenced by significant decreases in the number and time-varying probability of attack bouts, the total time spent attacking and the average duration of each attack bout; in addition, the latency to the 1st attack bout was significantly increased
Results, Oxtr/Avpr1a-mediated neuropeptidergic signaling in VMH is required for male territorial aggression, paragraph 3; Figure 2G, S2B-CHow it will be judged
Per mouse, over the first 15 min of the male-male assay (STAR Methods, Behavior annotations), the number of attack bouts, total attack time, mean attack bout duration and latency to the first attack, compared between groups with a Mann-Whitney test. Reproduced when each measure differs in the stated direction with p < 0.05; partially reproduced when some do.
- fig2-bilateral-editing-spares-sniff-and-mating
Bilateral co-editing does not change time spent sniffing male intruders, nor sniffing or mounting of female intruders.
the time spent in close investigation (sniffing) of male intruders did not differ significantly between experimental and control male residents ... Experimental males also did not differ significantly from controls in their sniffing or mounting behavior towards female intruders
Results, Oxtr/Avpr1a-mediated neuropeptidergic signaling in VMH is required for male territorial aggression, paragraph 3; Figure 2G-H, S2DHow it will be judged
A claim of absence, judged on the paper's test, the Mann-Whitney test between groups per mouse (n=11 each) on male-directed sniff time over the first 15 min of the male-male assay, and female-directed sniff time and the mounting measures of S2D over 10 min of male-female interaction (STAR Methods, Behavior annotations); for the inter-mount interval, mice with at most 1 mount bout are left out (same section). The percentage of animals mounting (S2Di) is a group proportion and is reported without a per-mouse test. Reproduced when none gives p < 0.05; the effect sizes are reported either way.
- fig3c-cre-editing-reduces-aggression
Bilateral Cre-dependent co-editing in VMHvl Esr1 neurons reduces most measures of aggression towards males, less strongly than the constitutive system, while sniffing duration is unaffected.
Bilateral co-disruption of Oxtr/Avpr1a in VMHvl Esr1 neurons significantly reduced most metrics of aggression towards male intruder compared to control mice, although to a somewhat lower extent than observed with the Cre-independent system
Results, Altered VMHvl Esr1 neural activity during social interactions, paragraph 2; Figure 3C, S3CHow it will be judged
Per mouse, over 15-20 min of male-male interaction, the attack measures of Figure 3C and S3C (percent attacking, attack bout duration, latency, inter-bout interval, number and probability of attack bouts) and sniff duration, Mann-Whitney between groups. Reproduced when more than half of the attack measures are significantly lower (or latency longer) at p < 0.05 and sniff duration is not significantly different.
- fig3fh-photometry-reduced
Bulk calcium activity of VMHvl Esr1 neurons during male-directed sniffing and attack is reduced in mice with bilateral Cre-dependent co-editing.
we observed a significant reduction in overall VMHvl Esr1 neuronal activity during the infrequent attack bouts exhibited by experimental mice ... A reduction in VMHvl Esr1 activity during sniffing episodes was also observed in these mice
Results, Altered VMHvl Esr1 neural activity during social interactions, paragraph 2; Figure 3E-HHow it will be judged
From the STAR Methods (Fiberphotometry recordings and analysis), the 470 nm signal corrected by a least-squares fit of the 405 nm signal, dF/F = (F470 - F405)/F405, normalized to a 2-s pre-bout baseline (mean and SD); AUC and peak over 10 s after bout onset, per bout; for sniffing, the first 1 min of sniffing after intruder entry. Mann-Whitney on bouts between groups. Reproduced when AUC and peak are lower in experimental mice at p < 0.05 for both sniffing and attack.
- figs3ef-photometry-female-unchanged
During interactions with females, bulk calcium activity during sniffing and mounting does not differ between experimental and control males.
during male-female interactions, calcium signals in experimental males were not significantly different from controls
Results, Altered VMHvl Esr1 neural activity during social interactions, paragraph 2; Figure S3E-FHow it will be judged
A claim of absence, judged on the paper's test, Mann-Whitney between groups on the AUC of the z-scored behavior-triggered average for female-directed sniffing (1 min) and mounting, computed as in fig3fh-photometry-reduced. Reproduced when neither gives p < 0.05.
- figs4a-unilateral-no-behavior-deficit
Mice with unilateral editing and imaging (the CRISPRoscopy mice) show no deficit in sniffing or aggression compared with controls.
Indeed, unilateral injected experimental animals displayed no deficits in sniffing or aggression compared to controls
Results, Single cell CRISPRoscopy imaging, paragraph 1; Figure S4AHow it will be judged
A claim of absence. The S4 legend names no test for panel A; the paper's test for the same measures in Figures 2, 3 and S3 is the Mann-Whitney test, used here. Per mouse, total sniff time, sniff bout duration, total attack time, attack bout duration, latency to first attack, inter-attack interval and attack probability, during the male-intruder period. Reproduced when none differs at p < 0.05.
- methods-unit-counts
CRISPRoscopy recorded 585 units from 5 control mice and 546 units from 7 experimental mice.
A total of 585 units (n=5 mice) from control and 546 units (n=7 mice) from experimental mice were recorded
STAR Methods, Micro-endoscopic data extraction, last sentenceHow it will be judged
Counts are exact; tolerance 0 because they are counts of recorded units and mice, not measurements. A difference is reported per mouse.
- fig4ef-male-activity-slightly-lower
During interactions with a male intruder, the activity of all VMHvl Esr1 units (sigma above pre-intruder baseline, and z-scored) is slightly but significantly lower in experimental than in control mice.
During interactions with a male intruder, the experimental cumulative distribution function (ECDF) and mean activity of all units (pooled from n=4 control and n=7 experimental animals) were slightly but significantly decreased in experimental mice
Results, Effect of Oxtr/Avpr1a co-editing on VMHvl Esr1 activity and intruder sex representations, paragraph 1; Figure 4Ei, 4Fi, S4DiHow it will be judged
Per unit, the mean activity in units of sigma above the pre-intruder baseline over the first part of the first male interaction, and the same for z-scored traces (z relative to the cell's mean over the whole recording). The window is open; the Results say "the first two minutes" and the Figure 4E-F legend "1 minute of interaction" (S4D also 1 min); readings 1 min and 2 min. Tests as named, under the file-level readings of "nested" and of the baseline. Reproduced when, under a reading, experimental activity is lower than control at p < 0.05 by both the Mann-Whitney and the KS test for sigma-normalized activity.
- fig4ef-female-activity-unchanged
During interactions with females, activity in units of sigma above baseline does not differ between experimental and control mice, while z-scored activity is slightly but significantly higher in experimental mice.
During interactions with females there was no significant difference in activity (measured in σ above baseline) between experimental and control animals ... although z-scored activity showed a slight but significant increase
Results, Effect of Oxtr/Avpr1a co-editing on VMHvl Esr1 activity and intruder sex representations, paragraph 1; Figure 4Eii, 4Fii, S4DiiHow it will be judged
As in fig4ef-male-activity-slightly-lower, for the first female interaction. The absence part is judged on the paper's tests (Mann-Whitney on mean activity, KS on the distribution, sigma-normalized). Reproduced when neither finds a difference at p < 0.05 and the KS test on z-scored activity finds experimental higher at p < 0.05. Under several readings of window and nesting, the outcome is the one they all give.
- figs4e-active-units-unchanged
Among male-responsive units (at least 2 sigma above baseline), mean activity does not differ between control and experimental mice.
However, the mean activity among cells considered as “active” (≥ 2σ above baseline 66 ) did not differ between control and experimental mice
Results, Effect of Oxtr/Avpr1a co-editing on VMHvl Esr1 activity and intruder sex representations, paragraph 1; Figure S4EHow it will be judged
A claim of absence. Units with mean activity of at least 2 sigma above the pre-intruder baseline during the male interaction window (readings as in fig4ef-male-activity-slightly-lower); nested Mann-Whitney between groups under the file-level readings. Reproduced when the test finds no difference at p < 0.05.
- fig4g-male-selective-reduced
The fraction of male-activated units (selective or mixed) is slightly smaller in experimental than in control mice (34.6% against 38.3%), and male-selective units are reduced by about 56% (6.3 +/- 2% against 14.4 +/- 4%).
The percentage of all male-activated neurons (selective or mixed) was slightly smaller in experimental than control mice (34.6% vs. 38.3%, respectively ... while there was a ~56% reduction in the small fraction of male-selective neurons (6.3 ± 2% vs. 14.4 ± 4%, respectively
Results, Effect of Oxtr/Avpr1a co-editing on VMHvl Esr1 activity and intruder sex representations, paragraph 2; Figure 4G, S4FHow it will be judged
A unit is male-selective when its activity is at least 2 sigma above the pre-intruder baseline during the male interaction and below 2 sigma during the female interaction, and mixed when at least 2 sigma during both (Results; Figure 4G legend). Window and baseline as in fig4ef-male-activity-slightly-lower. The percentages can be read as (a) pooled over units of all mice of the group, or (b) the mean over mice (the +/- values suggest b for the selective fractions). Reproduced when, under a reading, all four values are within tolerance; partially when the experimental male-selective fraction is lower than control but values fall outside.
- fig4g-female-selective-and-mixed-increased
During female interactions the fractions of female-selective and of mixed-selectivity units are increased by about 4% and 44% in experimental mice.
Conversely, during female interactions the fraction of female-selective and mixed selectivity neurons was increased by ~4% and ~44%, respectively, in experimental mice
Results, Effect of Oxtr/Avpr1a co-editing on VMHvl Esr1 activity and intruder sex representations, paragraph 2; Figure 4G, S4FHow it will be judged
Fractions defined as in fig4g-male-selective-reduced. "Increased by" can be read as (a) a relative change, (experimental - control) / control, or (b) a difference in percentage points; and the fractions as pooled or per-mouse means. Reproduced when, under one reading, both increases are within tolerance; partially when both fractions are higher in experimental mice but outside tolerance.
- fig4d-sex-decoding-intact
Linear SVM decoders trained on VMHvl Esr1 activity from either control or experimental mice predict intruder sex with virtually 100% accuracy.
linear SVM decoders trained on imaging data from either control or experimental mice correctly predicted intruder sex with virtually 100% accuracy
Results, Effect of Oxtr/Avpr1a co-editing on VMHvl Esr1 activity and intruder sex representations, paragraph 3; Figure 4DHow it will be judged
From the STAR Methods (Decoding intruder sex), per mouse, frame-wise linear SVM on N-unit population vectors from all frames of social interaction, equal numbers of male and female frames, stratified five-fold cross-validation, repeated 100 times, accuracy averaged per mouse; shuffle control with labels assigned at random to each behavior bout, 500 repeats. Reproduced when the mean accuracy of each group is at least 90% and above the shuffled accuracy at p < 0.05 (Mann-Whitney across mice). The SVM regularization is not given; readings C = 0.1, 1 and 10, and the outcome is the one they all give.
- fig4c-pls-sex-separation
Partial least-squares regression of population activity against intruder sex separates responses to males and females in both control and experimental mice.
This analysis revealed a clear separation of responses during encounters with males vs. females in both control and experimental animals
Results, Effect of Oxtr/Avpr1a co-editing on VMHvl Esr1 activity and intruder sex representations, paragraph 3; Figure 4C, S4B-CHow it will be judged
From the STAR Methods (Dimensionality reduction), per mouse, PLS of all unit traces against a vector that is -1 with a male intruder present, 1 with a female, 0 otherwise. The paper gives no measure of "clear separation". Readings of it (a) the mean PLS1 score differs in sign between male and female frames; (b) the area under the ROC curve of PLS1 scores, male against female frames, is above 0.75 or below 0.25. Reproduced when, under both readings, every mouse of both groups separates; blocked as methods_underspecified when the readings disagree. The PLS1 variance explained per group (S4C) is reported.
- fig5ac-behavior-activity-male-lower
Single-unit activity during male-directed sniffing and attack is slightly but significantly lower in experimental than in control mice, while activity during female-directed sniffing and mounting is similar.
The ECDF and average single unit activity during male-directed sniffing or attack was slightly but significantly lower in experimental than in control mice ... In contrast, during social interactions with females, activity during sniffing and mounting was similar
Results, Effect of Oxtr/Avpr1a co-editing on behavior representation in VMHvl, paragraph 1; Figure 5A, 5C, S5A-BHow it will be judged
Per unit, mean activity (sigma above pre-intruder baseline; z-scored in S5) over all frames annotated as each behavior. Tests as named under the file-level readings. Reproduced when male-directed sniffing and attack are both lower in experimental mice at p < 0.05 and, for female-directed sniffing and mounting, the paper's tests on sigma-normalized activity find no difference at p < 0.05.
- figs5c-peth-activity
Mean bout-triggered activity is slightly but significantly lower in experimental mice during male-directed sniffing and attack, higher during female-directed sniffing, and unchanged during mounting.
The mean activity during male-directed sniffing or attack was slightly but significantly lower in experimental than in control mice ... while it was significantly higher during female-directed sniffing, but unchanged during mounting
Results, Effect of Oxtr/Avpr1a co-editing on behavior representation in VMHvl, paragraph 1; Figure S5CHow it will be judged
From the STAR Methods (Micro-endoscopic data extraction), for each unit and bout, the z-scored trace minus the mean of a 2-3 s pre-bout baseline, averaged over 15 s after onset. The baseline length is open; readings 2 s and 3 s. Nested Mann-Whitney under the file-level readings. Reproduced when all four directions hold (male sniff and attack lower at p < 0.05, female sniff higher at p < 0.05, mounting not different at p < 0.05).
- fig5bd-behavior-selective-units-reduced
During male interactions, sniff-selective units are reduced from 8.9 +/- 3% to 3.9 +/- 2% and attack-selective units from 6 +/- 0.9% to 1.8 +/- 0.7% in experimental mice; the mixed-selectivity fraction is reduced by about 22%, and units active during male-directed behaviors fall from 36.2% to 22.4%; female-directed selective fractions are similar.
In experimental mice, during male interactions the fraction of sniff-selective units was reduced by ~40% relative to controls (3.9 ± 2% in OARs gRNAs mice vs. 8.9 ± 3% in Scr gRNAs mice) while the small proportion of attack-selective units was reduced by ~70% (1.8 ± 0.7% vs. 6 ± 0.9% ... The fraction of neurons exhibiting mixed behavioral selectivity (i.e., active during both behaviors) was moderately reduced (~22% ... Overall, there was a ~38% reduction in the fraction of active units during all male-directed behaviors (from 36.2% in control to 22.4% in experimental mice). In contrast, the fraction of female-directed sniffing- or mount-selective units was similar between control and experimental males
Results, Effect of Oxtr/Avpr1a co-editing on behavior representation in VMHvl, paragraph 2; Figure 5B, 5DHow it will be judged
Per unit, mean activity in sigma above the pre-intruder baseline over all frames of each behavior; selective for sniff when at least 2 sigma during sniffing and below 2 sigma during attack, and conversely; mixed when at least 2 sigma during both (Results; Figure 5B legend). "Active during all male-directed behaviors" read as (a) at least 2 sigma during sniffing or attack, or (b) the sum of the selective and mixed fractions as plotted in 5Di. Fractions as (a) pooled or (b) per-mouse means. Reproduced when, under one reading, all expected values are within tolerance; partially when the directions hold for sniff- and attack-selective and active fractions but some values fall outside. Female-directed selective fractions are reported with a Mann-Whitney test across mice; the text gives no test for them.
- fig5e-cp-tuning-shift
By choice probability, the percentages of attack- and sniff-tuned units during male-male interactions are reduced by about 80% and 70% in experimental mice, mixed-selectivity units increase by about 18%, and units tuned to sniffing females against sniffing males rise from 28.8 +/- 3% to 41.1 +/- 3%.
We observed an ~80% and ~70% reduction in the percentage of attack- and sniff-tuned units, respectively, in experimental vs. control mice during male-male interactions ... with a concomitant ~18% increase in the fraction of units exhibiting mixed selectivity ... In contrast, the percentage of units tuned to sniffing females (vs. sniffing males) was increased in experimental mice (28.8 ± 3% vs. 41.1 ± 3%
Results, Effect of Oxtr/Avpr1a co-editing on behavior representation in VMHvl, paragraph 3; Figure 5E, S5DHow it will be judged
From the STAR Methods (Choice probability), per unit and pair of behaviors, the area under the ROC curve built from histograms of the unit's dF/F values in frames of each behavior; shuffled 100 times by shuffling behavior bout timings; tuned when CP > 0.7 and more than 2 shuffled SDs above the shuffled mean (the Results print "p < 0.5"; the Methods' 2-SD rule is used). Open details and their readings. Direction, (a) CP(A vs B) > 0.7 means tuned to A and CP(B vs A) > 0.7 tuned to B, (b) only CP(A vs B) is computed and tuned to B means CP < 0.3 with the mirrored shuffle test. Bout shuffling, (a) circular shift of the bout sequence by a random offset, (b) random reassignment of bout onsets within the interaction period. The trace used for "dF/F", (a) the sigma-normalized trace, (b) the z-scored trace. Fractions as pooled or per-mouse means, and "reduction" or "increase" as relative change. For the sniff female against sniff male fractions, the values are assigned as the word "increased" requires (control 28.8%, experimental 41.1%); elsewhere in this section the paper lists the experimental value first, which would give the opposite assignment and contradict the text, so that reading is reported and does not decide. Every combination is run; reproduced when all the expected values are within tolerance under one combination and the direction of each change holds under all; blocked as methods_underspecified when the directions differ between combinations.
- figs5e-behavior-decoding-intact
Linear decoders distinguish attack from sniffing equally well in control and experimental mice.
there was no significant difference between control and experimental mice in the performance of linear decoders trained to distinguish attack from sniffing based on VMHvl Esr1 activity
Results, Effect of Oxtr/Avpr1a co-editing on behavior representation in VMHvl, paragraph 3; Figure S5EHow it will be judged
A claim of absence, judged on the paper's test. From the STAR Methods (Decoding behavior from neural activity), per mouse, frame-wise linear SVM on attack against sniffing frames towards male intruders, equal numbers of frames of each, 20 repeats, accuracy averaged per mouse; the cross-validation scheme is not given for this decoder, and the stratified five-fold scheme of the sex decoder is used as the paper's nearest stated choice. The Methods say frames are "sampled at 15 Hz" while traces are at 10 Hz; the data's own rate is used. Reproduced when the Mann-Whitney test across mice finds no difference between groups at p < 0.05; both groups' accuracy against shuffled decoders is reported.
- fig6-rslds-fit-quality
rSLDS models with 5 latent dimensions and 3 discrete states, fit to each mouse with at least 31 units, capture about 85% of the variance, with similar performance in control and experimental mice; attack occurs within a single rSLDS state in both groups.
In both the control and experimental groups, the fit models reduced the dimensionality of the data to 5 latent factors and three states (S1-S3) capturing ~85% of the observed variance in neural activity. Model performance (cvR 2) was similar between control and experimental mice ... During male-male interactions, attack behavior occurred during a single rSLDS state in both control and experimental males
Results, VMHvl Esr1 line attractor dynamics require Oxtr/Avpr1a-mediated signaling, paragraph 2; Figure S6DHow it will be judged
The rSLDS settings for all claims below are these. From the Results, one model per mouse with at least 31 units, 5 latent dimensions, 3 discrete states. From the ssm notebook the paper's code links, transitions recurrent_only, dynamics diagonal_gaussian, emissions gaussian_orthog, single subspace, Laplace-EM with a structured mean-field posterior, 100 iterations. Time bins of 0.1 s (traces downsampled to 10 Hz, STAR Methods). Open details and their readings. Which frames are fit, (a) the whole male-intruder part of the session from intruder entry to removal plus the post-intruder period, (b) the whole session (STAR Methods say the model relates "the activity of recorded neurons (y) during male directed behavior (sniffing, dominant mounting and attack)"). The trace, (a) z-scored, (b) sigma-normalized. Seeds and restarts are not given; fits are run with seeds 0 (the DJALab notebook's npr.seed(0)) to 3, and each claim's outcome is the one all seeds give. Performance is the forward-simulation score of the STAR Methods (normalized MSE of predictions over a short horizon, computed on held-out folds), with the horizon and folds not given (readings 1 s and 5 s; five folds over time); variance explained as the R2 of the emissions reconstruction. Reproduced when variance explained is within tolerance in both groups, performance does not differ at p < 0.05 (Mann-Whitney), and in each mouse attack "occurred during a single rSLDS state", read as (a) a majority of attack frames fall in one state, (b) at least 90% do.
- fig6a-control-tau1-greater
In control mice, the time constant of the first rSLDS dimension (about 100-120 s) is significantly larger than that of the second (about 40 s).
In control mice, the time constant (τ) of the 1 st rSLDS dimension (derived from the first eigenvalue of the fit dynamics matrix; see Methods ) was significantly higher than that of the 2 nd dimension (~100–120s vs. ~40s
Results, VMHvl Esr1 line attractor dynamics require Oxtr/Avpr1a-mediated signaling, paragraph 2; Figure 6AHow it will be judged
tau = |1 / log(|lambda|)| for each eigenvalue lambda of the dynamics matrix, in time bins, times 0.1 s (STAR Methods, Estimation of time constants), sorted in decreasing order. Which dynamics matrix is open; readings (a) the matrix of the discrete state in which attack occurs, (b) the mean of tau over the three states' matrices. rSLDS settings and readings as in fig6-rslds-fit-quality. Tau 1 reproduces inside the reported range widened by 10% of each bound (90 to 132 s); tau 2 within 10% of 40 s. Reproduced when, under a reading, the group means are within tolerance and the Kruskal-Wallis test across the five dimensions' tau values (mice as samples) gives p < 0.05 with dimension 1 the largest. The Figure 6 legend names no pairwise test after the Kruskal-Wallis test; readings (a) the Kruskal-Wallis test alone, (b) Dunn's multiple comparison of dimension 1 with dimension 2, the correction the paper pairs with Kruskal-Wallis in the S1F and S7 legends. With 4 mice a pairwise test may be unable to reach p < 0.05; the attainable minimum p is reported.
- fig6c-experimental-tau-indistinguishable
In experimental mice the first two rSLDS dimensions have statistically indistinguishable time constants (below 50 s), because the first dimension's time constant is reduced relative to controls.
In contrast, in experimental mice the first two rSLDS dimensions had statistically indistinguishable τ values (<50s; Figure 6C ) due to a reduced 1 st dimension τ
Results, VMHvl Esr1 line attractor dynamics require Oxtr/Avpr1a-mediated signaling, paragraph 2; Figure 6C, S6EHow it will be judged
Tau and readings as in fig6a-control-tau1-greater. Reproduced when, under a reading, the experimental mean tau 1 and tau 2 are below 55 s (50 s plus 10%), the test of tau 1 against tau 2 in experimental mice finds no difference at p < 0.05 (the absence is judged on the paper's Kruskal-Wallis test), and tau 1 is lower in experimental than in control mice at p < 0.01 by Mann-Whitney (the S6E image shows two stars).
- fig6f-line-attractor-score
The line attractor score, log2 of the ratio of the two largest time constants, is about 1.6 in control mice and close to zero in experimental mice.
This yielded a line attractor score (calculated as the log 2 of the ratio of the τ ‘s of the 1 st and 2 nd dimensions) of ~1.6 ... and consequently a line attractor score close to zero
Results, VMHvl Esr1 line attractor dynamics require Oxtr/Avpr1a-mediated signaling, paragraph 2; Figure 6FHow it will be judged
Score = log2(tau_n / tau_n-1), the two largest time constants (STAR Methods), per mouse; tau and readings as in fig6a-control-tau1-greater. "Close to zero" is judged by the Methods' own reading, a score below 1 (the Methods say it "is greater than one for systems that possess a line attractor"). Reproduced when, under a reading, the control mean is within tolerance, the experimental mean is below 1, and Mann-Whitney finds control higher at p < 0.05.
- figs6f-mating-attractor-intact
During mounting of female intruders, co-editing does not significantly reduce the first-dimension time constant or the line attractor score.
during mounting behavior towards female intruders, co-perturbation of Oxtr/Avpr1a-mediated signaling did not significantly reduce the 1st dimension τ values or the line attractor score in experimental mice compared to controls
Results, VMHvl Esr1 line attractor dynamics require Oxtr/Avpr1a-mediated signaling, paragraph 2; Figure S6FHow it will be judged
A claim of absence, judged on the paper's Mann-Whitney test. rSLDS fit to the female-intruder part of each session (readings for the frames as in fig6-rslds-fit-quality, with female-directed behavior in place of male-directed), mice with at least 31 units; tau 1 and score as in fig6a-control-tau1-greater and fig6f-line-attractor-score, with the state in which mounting occurs in place of attack for reading (a). Reproduced when neither tau 1 nor the score is lower in experimental mice at p < 0.05.
- fig6g-flow-field-line-vs-point
Flow fields of the fit rSLDS in the plane of the first two principal components show a line attractor (a roughly linear region of low flow) in control mice and a point attractor in experimental mice.
The 2D flow field of control mice (Scr gRNAs) revealed a roughly linear region of low vector flow constituting the line attractor ... In contrast, in experimental mice (OARs gRNAs) this line attractor was absent and was replaced by a point attractor
Results, VMHvl Esr1 line attractor dynamics require Oxtr/Avpr1a-mediated signaling, paragraph 3; Figure 6G-H, S6CHow it will be judged
The flow field is drawn from the fit dynamics as in the DJALab notebook (dx = A x + b on a grid, speed = norm of dx, the 3D landscape its height), in the plane of the first two principal components of the latent trajectories. The paper gives no measure of the shape of the low-flow region; its only measure of a line against a point attractor is the line attractor score (STAR Methods, score above 1 for a line attractor, near 0 for a point attractor). Judged per mouse, reproduced when the score is above 1 in a majority of control mice and below 1 in a majority of experimental mice. The flow field and landscape of every mouse are shown and described, and do not decide the outcome.
- fig6e-integration-dimension-persistence
In control mice the activity of the first rSLDS dimension ramps up during sniffing and attack and stays elevated after the intruder is removed; in experimental mice it decays rapidly between attack bouts and is significantly lower in the post-intruder interval.
in control mice this activity ramped up during the progression from male-directed sniffing to attack, eventually reaching a plateau where it decayed slowly between attack bouts towards a single intruder and remained elevated between sequential trials with different intruders ... In contrast, in experimental mice 1 st dimension neural activity decayed rapidly between attack bouts, displaying a “sawtooth” profile, and was significantly lower during the post-intruder (i.e., inter-trial) interval
Results, VMHvl Esr1 line attractor dynamics require Oxtr/Avpr1a-mediated signaling, paragraph 4; Figure 6B, 6D, 6E, S6A-BHow it will be judged
The weighted average of unit activity on dimension 1 (the latent x1 of the fit, equivalently the emission-weighted projection), normalized to 0-1 per mouse; rSLDS readings as in fig6-rslds-fit-quality. Per mouse, its mean over all frames of male-directed behavior bouts and over the post-intruder period. The post-intruder window is not given; readings (a) from removal of the intruder to the next intruder's entry or the end of the male part, (b) the first 60 s after removal. Reproduced when, under a reading, the paired t-test across experimental mice finds post-intruder lower than during bouts at p < 0.05 and the same test across control mice does not; partially when the experimental drop is significant and the control drop is also significant but smaller.
- fig7b-achw-first-dimension
The mean autocorrelation half-width of units weighted on the first rSLDS dimension, over the entire male-male interaction, is shorter in experimental (9.86 +/- 1 s) than in control mice (28.6 +/- 1.23 s); the reduction is about 66%, against about 61% for second-dimension units.
The mean ACHW of these neurons across the entire social interaction was significantly shorter in experimental (9.86 ± 1 s) than in control mice (28.6 ± 1.23 s), by ~20 seconds ... A reduction in the average ACHW was also observed among 2 nd rSLDS dimension cells
Results, Oxtr/Avp1ra-mediating signaling controls VMHvl Esr1 persistent neural activity, paragraph 1; Figure 7B, S7AHow it will be judged
ACHW per unit is the lag at which the unit's autocorrelation first falls to half its zero-lag value, in seconds (an "approximate measure of the decay constant", Results; the computation is cited to Remedios et al. 2017 and not given here). Open details and their readings. The trace, (a) z-scored, (b) sigma-normalized. Which units are "weighted on" a dimension, (a) each unit assigned to the dimension of its largest absolute emission weight, (b) the units in the top 25% by absolute weight on that dimension (the Results' "~20–25%" of units). "Average" ACHW per unit, (a) one autocorrelation over the whole male-intruder period, (b) the mean over intruder trials. rSLDS readings as in fig6-rslds-fit-quality. Every combination is run; reproduced when, under one, the three expected values are within tolerance and the paper's tests (under the file-level readings of "nested") give p < 0.05 with experimental shorter; blocked as methods_underspecified if combinations disagree on the direction or significance.
- fig7c-achw-all-units
Across all VMHvl Esr1 units, the mean ACHW during the first 1-2 minutes of male-male interaction is reduced in experimental mice (by about 50.9%), less than for first-dimension units.
A reduction in the average ACHW was also observed among 2 nd rSLDS dimension cells, as well as in the VMHvl Esr1 population as a whole ... the difference in mean ACHW between experimental and control animals was larger for 1 st dimension-weighted units (~66%) than for 2 nd dimension and total Esr1 + neurons (~61% and 50.9%, respectively
Results, Oxtr/Avp1ra-mediating signaling controls VMHvl Esr1 persistent neural activity, paragraph 1; Figure 7CHow it will be judged
ACHW and its readings as in fig7b-achw-first-dimension, all units of each mouse, over the first part of the first male interaction; the window is "the first 1–2 minutes" (Figure 7C legend), readings 1 min and 2 min. Reproduced when, under a reading, the reduction is within tolerance and the experimental mean is lower at p < 0.05 by the paper's tests.
- figs7b-achw-female-slight
During male-female interactions, the mean ACHW of all VMHvl Esr1 units is only slightly reduced in experimental mice (by about 22%).
In contrast, the average and cumulative distribution of ACHWs among all VMHvl Esr1 neurons were only slightly reduced (by ~22%) during male-female interactions
Results, Oxtr/Avp1ra-mediating signaling controls VMHvl Esr1 persistent neural activity, paragraph 1; Figure S7BHow it will be judged
As fig7c-achw-all-units, for the first 1-2 minutes of the first female interaction (S7B legend). The S7 legend names no test for panel B; the tests of the matching male panel (Figure 7C, nested Mann-Whitney on the mean, Kolmogorov-Smirnov on the distribution) are run and reported, and do not decide the outcome. The S7B image is read as showing one and two stars, and p < 0.05 is recorded. Reproduced when, under a reading, the reduction is within tolerance.
- figs7cd-slice-rslds
An rSLDS fit to slice calcium responses to OXT plus AVP captures 85% of the variance, with a first-dimension time constant (about 90 s) 8-9 times that of the second (about 15 s).
This fit model captured 85% of the observed variance in neural activity. The time constant of dimension x1 (~90s) was ~8–9 -fold greater than that of x2 (~15s
Results, OXT and AVP evoke persistent responses in VMHvl Esr1 neurons ex vivo, paragraph 1; Figure S7C-DHow it will be judged
rSLDS settings as in fig6-rslds-fit-quality; the number of latent dimensions and states for the slice fit is not given (Figure S7D shows two dimensions), and the frame rate is about 1.2 Hz (STAR Methods). Readings (a) 2 latent dimensions, (b) 5 latent dimensions with the two largest tau reported; the number of states read as 1 and 3. The paper's numbers are not consistent with each other (90 s over 15 s is 6, not 8-9), so the fold ratio is not an expected value; the ratio found is reported. Reproduced when, under one reading, all three values are within tolerance.
- figs7k-persistent-and-transient-slice-responses
Slice responses to OXT plus AVP fall into persistent responses that outlast the perfusion window and transient responses that decay within it.
Plotting the time-varying weighted average activity of x1 and x2 neurons revealed that the former exhibited slowly decaying responses to OXT+AVP application, while the latter exhibited more transient responses
Results, OXT and AVP evoke persistent responses in VMHvl Esr1 neurons ex vivo, paragraph 1; Figure S7F-KHow it will be judged
From the Figure S7K legend, a cell is transient when its response decays to 20% of the peak within the peptide perfusion window and persistent when it decays to 20% after the window. Reproduced when both classes contain cells and the x1-weighted cells (largest absolute weight on x1) are mostly persistent while x2-weighted cells are mostly transient.
- fig3d-cre-editing-spares-mounting
Bilateral Cre-dependent co-editing in VMHvl Esr1 neurons does not change mounting of female intruders.
consistent with the lack of an effect on mounting behavior
Results, Altered VMHvl Esr1 neural activity during social interactions, paragraph 2; Figure 3D, S3DHow it will be judged
A claim of absence, judged on the paper's test, the Mann-Whitney test between groups per mouse (Figure 3 legend) on the measures of Figure 3D and S3D (mean mount bout duration, latency to the first mount, interval between mount bouts) over 10 min of male-female interaction (STAR Methods, Behavior annotations); for the interval, mice with at most 1 mount bout are left out (same section). Reproduced when none gives p < 0.05; the effect sizes are reported either way.
- figs7l-blockers-slight-effect
In slices, synaptic transmission blockers (CNQX and MK-801) only slightly increase the decay rate of the peptide-evoked calcium response, during its later phases.
the inclusion of synaptic transmission blockers (20 μ M CNQX and 10 μ M MK-801) only slightly increased the decay rate during later phases of the peptide response
Results, OXT and AVP evoke persistent responses in VMHvl Esr1 neurons ex vivo, paragraph 1; Figure S7LHow it will be judged
From the S7L legend, the mean response of 45 cells to 100 nM AVP with 20 uM CNQX and 10 uM MK-801. The S7 legend names a Kruskal-Wallis test with Dunn's correction over time points 0-1000 s, and the text does not say the effect is significant, so no p-value is recorded; the test is run and reported. The comparison condition is not given (readings, the response to AVP without blockers, or to OXT plus AVP without blockers), nor which time points are "later phases". Reproduced when, under a reading, the blocker response decays faster than the comparison after its peak and not before it.
Pipeline
- Paper
- done (1 of 1)
- Claims
- done
- Claims check
- done (ready)
- Attempt
- todo
- Questions
- none
- Verified
- none