Rapid dynamics of electrophysiological connectome states are heritable

Figures

• 

  Figure 1.. An overview of the analysis pipeline. [A] We used resting-state source-space EEG and structural images from Minnesota Twin Family Study samples. We employed a hidden Markov model (HMM) to extract K=6 discrete connectome states (or K=4 states for replication) for each canonical EEG frequency band associated with a state time course for each subject indicating the probability of when each state is active. The amplitude coupling-based FC matrices of the states are shown at the bottom. The six states are color coded (blue, red, yellow, green, purple, and light blue) to illustrate their contribution to the connectome’s dynamic features of interest. [B] We constructed each feature in a multivariate manner to comprehensively represent all states. Specifically, multivariate temporal features were defined as the proportion of the recording time spent in each connectome state (Fractional Occupancy) and the probability matrix of transitioning between all possible pairs of discrete states (Transition Probability). Multivariate spatial features include time-varying Modularity (ModularityTime-Varying), and time-varying connectivity strength (FCTime-Varying) averaged across the set of…

• 

  Figure 2.. HMM states and state-dissociating features. (A) From band-specific leakage-corrected EEG signals (amplitude envelopes) concatenated over all subjects, HMM estimates connectome states that each have a characteristic FC matrix. The FC matrices reflect amplitude coupling among all region pairs. Here, the connectome states’ FC matrices for the alpha band (8–12 Hz) are provided as examples (See full visualization of the states for all canonical bands in Figure S4). The rows and columns represent 54 regions organized according to their membership to canonical intrinsic connectivity networks (ICNs listed on the left) (Yeo et al., 2011). (B) HMM estimates a specific (probabilistic) state time course for each subject indicating when each state is active. An approximately two-minute section of the state time course is visualized for one subject exemplifying periods occupied by each state and the transitions across states. (C) The binary matrix shows data-driven clusters of connections whose FC strength differed significantly across the six states. Specifically, F values from a connection-wise ANCOVA of the factor state were threshold at 5% connection density and entered Network-Based…

• 

  Figure 3.. Heritability of temporal features of the dynamic connectome. (A) Heritability of each of the multivariate connectome dynamics features was assessed separately by one-way ANCOVAs of the factor sibling status (three levels: monozygotic twins (MZ), sex-matched dizygotic twins (DZ), and sex-matched pairs of unrelated individuals), adjusted for age and sex. The main effect of sibling status indicates the heritability, or genetic effect. Bar graphs show that genetically more similar subject pairs have shorter Euclidean distance, indicating higher similarity of a given temporal feature. This effect of sibling status was large and independent of the chosen number of states (see Figure S5). P†: P values Bonferroni-corrected for 20 tests (four multivariate features and five frequency bands), η2p: Partial Eta squared effect size. (B) A visual illustration of the effects exemplified for two subject pairs, one from the MZ group (left) and the other from the unrelated subject group (right). While the MZ twins display high similarity of Fractional Occupancy and Transition Probabilities, both temporal features are dissimilar across the unrelated pair.