The thesis focus on the modularity of metabolic network and foremost on the architecture of regulatory network representing direct regulatory interactions between metabolites and enzymes. I focus on the "modularity measure" in my first work. Modularity measure is quantitative measure of network modularity commonly used for module identification. It was showed that algorithms using this measure can produce modules that are composed of two clearly pronounced sub-modules. Maximum size of module for which there is a risk that is is composed of two sub-modules is called resolution limit of modularity measure. In my first work I generalize resolution limit of modularity measure. The generalized version provide insight to the origin of resolution limit in the null-model used by modularity measure. Moreover it is showed that the risk of omitting of sub-modular structures applies for bigger modules than mentioned in the original publication. The second work is focused on the question how does the modular structure of E. coli metabolic network change if we add regulatory interactions. I find that the modularity of modular core of network slightly increase after regulatory edges addition. The modularity increase is significant with respect to randomized ensemble of regulatory networks. Identified modules...