Asymmetry in modular networks.

<p>(A) Activity intensity of 100 consecutive NBs for each of the clusters in a three cluster chain. Inset: a schematic drawing of the network under investigation. (B) Reordered similarity matrix (see <a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1004883#sec013" target="_blank">Materials and Methods</a> for details) between 1000 consecutive NBs in this cluster chain. Only NBs in more than one cluster were considered. A clear separation of NB patterns into similarity groups is seen. (C) Average NB activity patterns are marked by the solid line rectangles in (B). Pattern number 2 was divided into two groups (2a and 2b, dashed lines) to exemplify variability within a group. (D) A schematic representation of the functional asymmetry calculation. Long-term asymmetry is calculated from the cross-correlations function as the integral over its positive side subtracted from its negative side and divided by the sum of both. (E) Distribution of asymmetry values between pairs of clusters (n = 48) in which only one cluster (marked as c2 in the inset) is connected to additional clusters (marked as x in the inset). c2 is connected to the rest of the network through one link (white bars), or more than one link (black bars). Negative asymmetry corresponds to activity propagation from c2 to c1. Inset: Network schematics. (F) Activation delays as function of time between two specific coupled clusters. Delays were calculated from the location of the cross-correlation function peak, for consecutive NBs during a ten-hour recording (only NBs activating both clusters were analyzed). The black line represents the average delay in one-hour windows. Color coding represents the asymmetry for consecutive NBs.