What a bacterial cell count leaves out

Imagine six bacterial cells. Three isolated pairs and a single chain of six have the same headcount. Their organisation differs: in the chain, a sequence of links can join every cell to every other; the pairs remain separate.

Two neural networks turn electron-microscope images into graphs: one marks cells; the other identifies visible associations. Cells become nodes; associations become edges. A scale-adjusted distance filter excludes distant pairs.[1]

Count the connections too

A connected component contains nodes that can reach one another along a path of edges. No such path leads to a different component. This definition, described in NetworkX’s graph documentation, lets an analysis distinguish one connected assembly from several separate groups.[2]

Joining two pairs needs just one additional link. The headcount stays four, but two components become one. A component count is therefore sensitive to the placement of links, while an edge count measures how many exist. Looking at both preserves information that either number alone would lose.

The researchers studied Pseudomonas aeruginosa on polylactide and polycarbonate over 2–22 hours. In one stitched field of view, the base graph contained 822 cells and 392 edges, forming 461 components. Its largest component contained 22 cells.[1]

What an edge can establish

An edge represents visible association on the imaged surface, such as contact. It does not identify chemical communication. Mature biofilms obscure cell boundaries, so the method addresses earlier development.[1]

Those connections make the organisation of an early biofilm available for measurement.[1]

Report an error or suggest a correction