In bacterial genomes a fraction of transcribed sequences do not code for proteins or structural RNAs, but have been shown to be involved in fundamental processes, such as regulation of gene expression, mRNA processing and stability or structural RNA maturation. In this thesis a systematic procedure to identify and classify families of repeated sequences sharing a common RNA secondary structure was applied to the study of 40 bacterial genomes. Sequences able to fold in a stable stem loop structure were clustered according to sequence similarity, and grouped within homogeneous families. The study led to the identification of 57 families of repeated sequences, sharing a common secondary structure and potentially coding for structured RNAs. All previously known such families have been detected by the used procedure, and are listed within the final set, together with 37 novel ones. Their location in relation to protein coding genes was evaluated, and a correlation was found between structure and positioning within intergenic regions. A new software tool is also described, Scaffolder, designed to help in high-throughput de novo genome sequencing by finding connections between contigs produced by random shotgun sequencing, and assisting the researcher in the whole process. The software, accessible both as a command line tool and as a web application, can guide all the final phases of genome assembly by storing the current assembly status, displaying networks of connected contigs and untangling multiply connected ones by a combination of computational and experimental procedures.