We define a model which builds upon the combination of two popular tools to infer gene regulatory networks from transcriptomic data, namely Gaussian Graphical Models (GGMs) and l1 regularization. GGMs describe the graph of conditional dependencies between genes while l1 regularization helps solve the problem under high-dimensional settings and select of the set of relevant interactions.

Ambroise et al 2009 suggested to improve the selection of edges thanks to the fine modeling of the network structure. Indeed, gene regulatory networks are known not only to be sparse, but also organized, so as genes belong to different classes of connectivity. It thus seems intuitive to search for regulations preferentially between genes where a prior structure suggests they should be.

In this presentation, we extend this approach to VAR1 modeling in order to be able to handle time series datasets, understood as one single campaign of repeated measurements over time.

References:
Ambroise, C., Chiquet, J., & Matias, C. (2009). Inferring sparse Gaussian graphical models with latent structure. Electronic Journal of Statistics, 3, 205-238
Charbonnier, C., Chiquet, J., & Ambroise, C. (2010). Weighted-Lasso for Structured Network Inference from Time Course Data. Statistical Applications in Genetics and Molecular Biology, 9.