Omics data have provided the scientific community with an unprecedented amount of information that calls for powerful analytical tools. Not only these tools should be able to unveil the most salient features of the data, but modeling should also feed on these features to delineate possible mechanisms and next-generation experiments.

In this talk, I will present a generic framework to extract multi-scale networks of functional dependencies. I will explain how these networks can be extracted from a possibly high level of background correlations (e.g. phylogenetic correlations) and how modeling frameworks can rationalize their properties. As an example, I will present recent results about the evolution of bacterial genomes, more particularly on the organizational aspects. In order to highlight the generality of the approach, I will discuss as well transcriptomics data in model organisms.