Comparative genomics is as a fundamental discipline to unravel evolutionary biology. To overcome a mere descriptive knowledge of it the first challenge is to develop a higher-level description of the content of a genome. Therefore we used the modular representation of genomes to explore quantitative laws that regulate how genomes are built from elementary functional and evolutionary ingredients. The first part sets off from the observation that the number of domains sharing the same function increases as a power law of the genome size. Since functional categories are aggregates of domain families, we asked how the abundance of domains performing a specific function emerges from evolutionary moves at the family level. We found that domain families are also characterized by family-dependent scaling laws. The second chapter provides a theoretical framework for the emergence of shared components from dependency in empirical component systems with non-binary abundances. We defined a positive model that builds a realization from a set of components linked in a dependency network. The ensemble of resulting realizations reproduces both the distribution of shared components and the law for the growth of the number of distinct families with genome size. The last chapter extends the component systems approach to microbial ecosystems. Using our findings about families scaling laws, we analyzed how the abundance of domain families in a metagenome is affected by the constraint of power-law scaling of family abundance in individual genomes. The result is the definition of an observable, whose functional form contains quantitative information on the original composition of the metagenome.