Endosymbiosis, a process by which one organism evolves to live within another, led to major evolutionary leaps. Studying old endosymbiosis is necessary for unraveling evolutionary history of life, host-adaptation strategies, phagocytosis and its role in eukaryogenesis. We propose to focus on Deep-branching Gammaproteobacteria (DIG) which are the oldest endosymbiont group and represents the Legionellales adapted to all hosts and several related intracellular groups (Francisellaceae, Fastidiobacteraceae and Piscirickettsiaceae). Moreover, this large group has evolved a wide variety of host-adaptation strategies, and its last common ancestor is estimated to be >2.0 Ga, likely coincident with the eukaryogenesis. In this project, we try to establish the evolutionary history of DIG by identifying novel genomes belonging to these ones. Given their intracellular lifestyle, it is difficult to isolate them and we use metagenomics and binning to recreate genomes. First, we are screening existing metagenomes, using a pipeline created in our lab. Briefly, we use a step-wise, sieve-like approach to target the computational power of the most promising metagenomes. We first identify and download metagenomes containing at least one read similar to a 16S rRNA gene attributed to DIG and select reads by the presence of specific marker genes. We then assemble the most promising metagenomes and screen theme by finding contigs containing genes attributed to the host-adaptation system. We also use phylogenomics based on these contigs to sub-select most promising assembled metagenomes. Finally, we perform binning on the remaining sequences. By identifying novel organisms belong DIG and analyzing their evolutionary history, we will better understand the relationship between the first eukaryotic hosts and their first intracellular invaders.