Genome-resolved metagenomics has led to the discovery of previously untapped biodiversity within the microbial world. As the development of computational methods for the recovery of genomes from metagenomes continues, existing strategies need to be evaluated and compared to eventually lead to standardized computational workflows.

Metagenome-Assembled Genomes (MAGs) represent individual genomes recovered from metagenomic data. MAGs are extremely useful to analyse uncultured microbial genomic diversity, as well as to characterize associated functional and metabolic potential in natural environments. Recent computational developments have considerably improved MAGs reconstruction but also emphasized several limitations, such as the non-binning of sequence regions with repetitions or distinct nucleotidic composition. Different assembly and binning strategies are often used, however, it still remains unclear which assembly strategy in combination with which binning approach, offers the best performance for MAGs recovery. Several workflows have been proposed in order to reconstruct MAGs, but users are usually limited to single-metagenome assembly or need to manually define sets of metagenomes to co-assemble prior to genome binning. Here, we present MAGNETO, an automated workflow dedicated to MAGs reconstruction, which includes a fully-automated co-assembly step informed by optimal clustering of metagenomic distances, and implements complementary genome binning strategies, for improving MAGs recovery. MAGNETO is implemented as a Snakemake workflow and is available at: https://gitlab.univ-nantes.fr/bird_pipeline_registry/magneto.IMPORTANCEGenome-resolved metagenomics has led to the discovery of previously untapped biodiversity within the microbial world. As the development of computational methods for the recovery of genomes from metagenomes continues, existing strategies need to be evaluated and compared to eventually lead to standardized computational workflows. In this study, we compared commonly used assembly and binning strategies and assessed their performance using both simulated and real metagenomic datasets. We propose a novel approach to automate co-assembly, avoiding the requirement for a priori knowledge to combine metagenomic information. The comparison against a previous co-assembly approach demonstrates a strong impact of this step on genome binning results, but also the benefits of informing co-assembly for improving the quality of recovered genomes. MAGNETO integrates complementary assembly-binning strategies to optimize genome reconstruction and provides a complete reads-to-genomes workflow for the growing microbiome research community.