Fig. 3: Unsupervised clustering analysis of fear acquisition data on tones 2–5.
A Cluster enrichment for the female fear acquisition data using tones 2–5. Bar graphs represent mean ± standard deviation of the time proportion spent on each cluster. Statistics were performed using an independent samples t-test corrected for multiple testing using Benjamini–Hochberg’s method across clusters between NS and LBN exposure. Significant differences were observed in clusters 0: T = −3.55, p = 2.28e−3, and 6: T = 4.24, p = 4.97e−4, but none of the other clusters (p > 0.05). B Cluster enrichment for the male fear acquisition data using tones 2–5. No significant differences were observed between NS and LBN mice using the independent samples t-test corrected for multiple testing using Benjamini–Hochberg’s method across clusters (p > 0.05). C The supervised cluster performance was performed via the validation performance per cluster across a 10-fold cross-validation loop. Balanced accuracy was used to correct for cluster assignment imbalance. The dashed line marks are the expected performance due to chance, considering all outputs. D The confusion matrix is obtained from the trained gradient boosting machine classifying between clusters. Aggregated performance over the validation folds of 10-fold cross-validation is shown. E The global SHAP feature importance between the different clusters. Features in the y-axis are sorted on the global absolute SHAP values across all clusters. The classes in the bar graphs are sorted by highest to lowest clusters importance within every feature. F, G Bee swarm plots for the two differentially expressed clusters within the female fear acquisition data between NS and LBN mice, clusters 0 and 6. The plots show the 10 most important features for each classifier, in terms of the mean absolute value of the SHAP values. Bar graphs represent mean ± standard deviation of the time proportion spent on each cluster. Panel A–F represent female NS (n = 10, female LBN (n = 10), male NS (n = 10), male LBN (n = 11).
