Fig. 5: CSDS damages eight-region brain network in social interaction.
A Global efficiency of Control and CSDS mice in the frequency bands. Ordinary Two-Way ANOVA with Šídák’s multiple comparisons test; a significant effect of group, F (1,270) = 32.11, P < 0.001. For low theta; CSDS: 0.0779 ± 0.0039, Control: 0.0793 ± 0.0030; P = 0.7024. For high theta; CSDS: 0.0644 ± 0.0022, Control: 0.0724 ± 0.0028; P = 0.0267. For beta; CSDS: 0.0734 ± 0.0027, Control: 0.0892 ± 0.0030; P < 0.001. For low gamma; CSDS: 0.0823 ± 0.0030, Control: 0.0950 ± 0.0025; P = 0.0018. For high theta; CSDS: 0.0864 ± 0.0029, Control: 0.0996 ± 0.0023; P = 0.0016. B Pearson correlation analysis between global efficiency and social index in high theta frequency band. Spearman correlation analysis. r = 0.3111, * P = 0.0196. C Pearson correlation analysis between global efficiency and social index for all frequency bands. The detailed correlation statistic results were shown in Supplemental Table 2. Black dotted line represents the significance level: 0.05. D Chord diagrams of information flow with significant change between Control and CSDS mice for individual frequency band. Different colors represent various brain regions. The arrows indicate the direction of information flow (IF), while the width of each arrow corresponds to the significance level of the change in IF between the CSDS and Control groups. E The contribution of individual frequency network in Control/CSDS classification. An ordinary one-way ANOVA with Tukey’s multiple comparisons test. F (4,495) = 21.19, P < 0.001. An ordinary one-way ANOVA with Tukey’s multiple comparisons test. F (4,495) = 21.19, P < 0.001. Slow theta: −5.75% ± 0.45%, Shigh theta: −2.80% ± 0.44%, Slow theta vs Shigh theta: P < 0.001; Slow gamma: −1.05% ± 0.36%, Slow theta vs Slow gamma: P < 0.001; Shigh gamma: −2.47% ± 0.39%, Slow theta vs Shigh gamma: P < 0.001. F Binary classification accuracy based on diverse learned features (oscillation, network and their combination). An ordinary one-way ANOVA with Tukey’s multiple comparisons test. F (2, 297) = 29.20, P < 0.001. Oscillation: 77.19% ± 0.28%, Network: 78.52% ± 0.28%, Both: 80.06% ± 0.24%. Oscillation vs Network: P = 0.0013; Oscillation vs Both: P < 0.001; Network vs Both: P < 0.001. G ROC curves of binary classification model based on different learned features (oscillation, network and their combination). H AUC of binary classification model based on different learned features (oscillation, network and their combination). An ordinary one-way ANOVA with Tukey’s multiple comparisons test. F (2, 297) = 296.5, P < 0.001. Oscillation: 0.7952 ± 0.0024, Network: 0.78581 ± 0.0025, Both: 0.8735 ± 0.0023. Oscillation vs Network: P < 0.001; Oscillation vs Both: P < 0.001; Network vs Both: P < 0.001. All data are shown as mean±s.e.m. Data were from Control group (n = 28 mice) and CSDS group (n = 28 mice). * P < 0.05, ** P < 0.01, *** P < 0.001.
