Last year, my collaborator Nicolas Städler and I developed a network analysis package for high-dimensional data. Now at version 1.0.1. (talk about incremental process), the nethet Bioconductor package could be of use to anyone whose day job involved working with large networks. The package is a catch-all containing a bunch of analysis and visualisation tools, but the two most interesting are:
- Statistical two-sample testing for network differences in graphical Gaussian models. This tool, which we call differential network (Städler and Mukherjee, 2013, 2015), allows us to test the null hypothesis that two datasets would give rise to the same network under a Gaussian model. Useful if you’re unsure whether you should combine your datasets or not.
- Combined network estimation and clustering using a mixture model with graphical lasso estimation (Friedman et al. 2008). We call this tool mixGLasso, and it allows for estimation of larger networks than would be possible using traditional mixture models. Think of it as network-based clustering, with the underlying networks determined by the inverse covariance matrix of the graphical Gaussian model. The tool will group together samples that have the same underlying network. Useful if you know your dataset is heterogeneous, but are not sure how to partition it.
Intrigued? You can download nethet using Bioconductor, or have a look at the vignette to see some worked examples of how to use it.