Flexible random graph modelling for cell–cell interactions

Abstract

In this paper we explore generalizations to a random digraph model designed for modelling cell–cell interactions (RaCInG). In its baseline model—tested both theoretically and in biological practice—a fixed number of arcs are assigned uniformly at random to vertices that have suitable types to accept them. Generalizations to this model are explored in two directions: constructing an undirected graph version of the model, and incorporating spatial information. Both directions are driven by biological demands: better interpretability and adaptation to technological advances. We show not only that these generalizations are theoretically possible within the framework of RaCInG, but that they can be made rigorous, proving that RaCInG creates a flexible random graph framework to model cell–cell interactions.