A graphical model is a class of statistical models that can be represented by a graph which can be used to identify conditional independence properties. Some common examples of graphical models are Bayesian networks (directed graphical models), log-linear models (undirected models), block-recursive graphical models, and models defined using the BUGS language. Today, there exists a wide range of packages to support the analysis of data using graphical models. Here, we focus on Open Source software, making it possible to extend the functionality by integrating these packages into more general tools.

We will attempt to give an overview of the available Open Source software, with focus on the gR project. This project was launched in 2002 to make facilities in R for graphical modelling. Several R packages have been developed within the gR project both for display and analysis of graphical models. This facilitates extensions in the form of R packages which may rely on the whole R system.

Examples of R packages in the gR project include: gRbase, defining a common data structure; giRaph, defining mathematical graphs; dynamicGraph, giving an interactive graphical user interface for manipulating graphs; CoCo, a bundle of algorithms for efficient analysis of graphical models; mimR, giving an interface to the MIM program; deal, learning parameters of a Bayesisan network; ggm, giving tools for Gaussian graphical models; BRugs, running BUGS within R; SIN, model selection in Gaussian graphical models.