The goal of janine is to perform network reconstruction by estimating the support of the precision matrix in a Gaussian graphical model, with the additional assumption that the underlying graph is organized according to a Stochastic Block Model (SBM). This SBM can possibly be multipartite when the target network concerns entities of different natures (like proteins and genes in genomics, or bacteria and fungi in microbial ecology), but associated to the same individuals (samples of patients or plant ecosystem for instance).
To adjust the model, with us weighted sparse technique possibly coupled with graph-Laplacian-guided regularization. The weigths and the graph-Laplacian are themselves estiamted with SBMs, in an iterative process.
Installation
You can install the development version of janine from github
remotes::install_github("jchiquet/janine")
Reference
Janine is a self private joke/reference to an older R package called simone, which has the same idea of driving network inference according to an underlying latent and unobserved organization of the network. We use a different rationale here which is more efficient and generalize to a wider class of SBM, in particular multipartite SBM.
