Fig. 4: Identify movement phenotypes on single experimental data.

a Spatiotemporal feature space of behavioral components. Each dot on the 3D scatter plot represents a movement bout (n = 935 bouts). The 11 different colors indicate the corresponding to 11 movement types. b Upper, recalculated paired-wise similarity matrix, and they were rearranged with a dendrogram (lower). Each pixel on the matrix represents the normalized similarity value of a pair of movement bouts at the ith row and the jth column. The color-coded bars indicate the labels of clustered movement (middle). c Fractions of movement bouts number. For each subject, the behavior fractions are defined as the bouts number of each behavioral phenotype divide by the total number of behavior bouts the animal occurred during the experiment. d Intra-CC (color-coded) and inter-CC (gray dots) of each movement group. The dots on each violin plot represents their intra-CC or inter-CC, and dots number in a pair of violin plot in each group are the same (Intra-CC: 0.91 ± 0.07; Inter-CC: 0.29 ± 0.19). e Cumulative Distribution Function (CDF) of CQI of the movement clusters. The clusters represented by the curves on the right side have better clustering qualities, and their corresponding movements are more stereotyped. f The histogram of the duration of all movements (0.963 ± 0.497 s). CC correlation coefficient, CDF cumulative distribution function, CQI Clustering Quality Index. Source data are provided as a Source Data file.