FIGURE 8.

Computational approach of the clustering method DBSCAN. (A): DBSCAN explanation. For every localization in a given FoV, the number of neighbours in a user-defined radius are counted. If there are more neighboring points than a (by the user) set minimal number of points (left), these points are considered core points (blue). All localizations that are not core points, but have a neighbour that is a core point, are considered an edge point (orange, second panel). Localizations that have neither enough neighbours nor proximal core localizations are considered noise points (red, third panel). The core and edge localizations together are considered to form a cluster. The identified clusters can be visualised and characterized (right). In this example, at least 3 neighbours are required for a localization to be considered a core point in the radius indicated by the circle. (B): Application of the DBSCAN steps on an E. coli cell with fluorescently-labeled RNA polymerase.