Numerous advantages of systematic genomics have been leveraged in other areas, such as transcriptomics, proteomics and human microbiome studies, but it has yet to be effective in stimulating the increased discovery of natural products (secondary metabolites).1 Since the discovery of the first microbial genome (2002), the surprisingly high percentage of cryptic biosynthetic pathways per genome (genes not connected to any known secondary metabolite, aka natural product) has continued as a consistent theme.2 Cyanobacteria from benthic environments are prolific producers of natural products but genomes from these organisms remain scarce in public repositories. Hence, this dissertation focused on sequencing and performing genomic analysis in prolific cyanobacteria collected from several tropical benthic ecosystems around the planet. First, we analyzed the relationship between a cyanobacterium host and its associated heterotrophic microbiome. By evaluation the non-axenic culture of Moorea producens JHB, we characterized a novel uncultured acidobacteria heterotroph living with the cyanobacterial host.3 This heterotroph was only 85% similar to the closest cultured representative, it presented a large number of genes encoding for transcriptional regulators and it was auxotrophic for several proteogenic amino acids. Next, we expanded our genome comparison by deeper analyzing 4 Moorea genomes. Our genome comparison revealed that Moorea, already a prolific producer of secondary metabolites, harbored an even richer metabolic potential, four times above the cyanobacterial average. We observed that Moorea conserved its primary metabolism while evolved an intricate secondary metabolism machinery, accounting for 20% of their genomic content. These findings were very promising for future genome mining efforts in those four strains. Thereby, we further expanded our sequencing efforts by sampling 165 metagenomes of filamentous marine cyanobacteria collected from around the globe. Our new metagenomic pipeline was able to generated 85 high-quality genomes, including 32 Moorea and 29 Okeania strains. Our genome comparison highlighted that these two genera are among the most diverse and prolific producers of natural products in our dataset (comparing pairwise 506 cyanobacterial genomes, 425 from the NCBI database). Gene networking revealed the abundance of unique (only encountered in a single strain) and “extended families” (found in several strains). Using genomes with paired mass spectrometry data, we correlated the distribution of gene families with the distribution of metabolic families and automatically annotated a new dehydromicrosclerodermin B homologue.