Figure 1

Overview of the Graphle system architecture. The Graphle server manages up to hundreds of gigabytes of weighted undirected graphs; while any graph data can be used, Graphle is specifically designed for biological networks in which vertices represent genes and edges represent experimental results (microarray correlations, protein-protein interactions, etc.) or computational predictions (e.g. probabilities of functional interactions). The server also associates metadata with graphs (such as what organism or biological context they are drawn from), vertices (gene identifiers, aliases, known cellular functions, etc.), and edges (e.g. what experiments or data contributed to that edge). The Graphle client communicates user-provided queries to the server consisting of one or more genes of interest, receives an appropriate subgraph, and displays it interactively for the user in a web browser. The user can then change the focused genes or the stringency cutoff for vertex or edge weights and can access the associated metadata to interactively explore tractable portions of the large underlying graphs.