Fig. 5: The identification of abnormal social behaviour modules in Shank3B KO mice.
a, The fractions of social behavioural modules of three social groups. The fractions of each group are normalized, and they are clustered and re-sorted according to the dimension of social behaviour modules. b, Dimensional reduction of behaviour fractions using PCA after hypothesis testing (two-way ANOVA followed by the Tukey multiple comparisons test). In the three groups, 24 social behaviour modules show significant differences. Three components can explain more than 90% variances, and 11 components can explain more than 99% variances. c, The construction of phenotype space. UMAP is used to reduce the 260 dimensions of social behaviour modules to three dimensions according to e. Different coloured dots represent different social groups. The phenotypes of three social groups can be separated in phenotype space. d, The merging of social behaviour modules according to behavioural feature angles and b. First, 24 social behaviour modules with significant differences are mapped to PCA feature space, and then the angular separation is calculated to construct the angle spectrum. Further, hierarchical clustering is used to cluster the angle spectrum into 11 clusters according to b. e, The comparison of behavioural fractions of three social groups: 24 social behaviour modules with significant differences are manually identified (mean ± s.d., two-way ANOVA followed by Tukey multiple comparisons test, n = 20, adjusted P values from left to right (group A versus group B, group A versus group C and group B versus group C) are >0.9999, <0.0001, <0.0001, 0.9990, <0.0001, <0.0001, >0.9999, <0.0001, <0.0001, 0.9939, 0.0002, <0.0001, >0.9999, <0.0001, <0.0001, 0.9919, 0.0016, 0.0010, 0.8210, 0.0055, 0.0331, 0.0001, 0.0029, 0.7179, 0.8213, 0.0438, 0.1703, 0.9882, <0.0001, <0.0001, 0.2691, 0.0323, 0.6000, 0.5677, 0.0952, 0.0057, 0.6034, 0.0101, 0.1239, 0.0733, 0.0145, 0.8183, 0.2698, 0.0184, 0.4735, 0.1217, 0.0011, 0.2511, 0.4016, 0.0397, 0.4864, 0.5691, 0.0001, 0.0054, 0.0005, <0.0001, <0.0001, 0.6728, 0.0297, 0.2076, 0.0175, 0.7233, 0.1220, 0.0445, 0.2145, 0.7555, 0.0222, 0.4986, 0.2810, 0.1373, 0.8823 and 0.0454). f, The visualization of merged social behaviour modules. With the assistance of d, nine social behaviour modules are merged and identified from 24 social behaviour modules. The colour of mice represents the behaviour cases with the highest mean fraction in e. The orange 3D mice represent KO mice and green 3D mice represent WT mice. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.
