During the course of evolution, structurally autonomous modules of old proteins are frequently reused and recombined into new proteins. These modules are known as protein domains. This thesis deals with the discovery of previously unknown protein domains based on sequence information alone. Five single discoveries of novel protein domains are described in detail. Furthermore, two applications of models of protein domains in the analysis of completely sequenced genomes are presented. The analyses provided valuable new insights into different aspects of molecular cell biology, in particular the evolution of the nucleolus, and in human hereditary diseases, i.e. autoinflammatory syndromes and epilepsy.