BackgroundMetagenomic sequencing has lead to the recovery of previously unexplored microbial genomes. In this sense, short-reads sequencing platforms often result in highly fragmented metagenomes, thus complicating downstream analyses. Third generation sequencing technologies, such as MinION, could lead to more contiguous assemblies due to their ability to generate long reads. Nevertheless, there is a lack of studies evaluating the suitability of the available assembly tools for this new type of data.FindingsWe benchmarked the ability of different short-reads and long-reads tools to assembly two different commercially available mock communities, and observed remarkable differences in the resulting assemblies depending on the software of choice. Short-reads metagenomic assemblers proved unsuitable for MinION data. Among the long-reads assemblers tested, Flye and Canu were the only ones performing well in all the datasets. These tools were able to retrieve complete individual genomes directly from the metagenome, and assembled a bacterial genome in only two contigs in the best scenario. Despite the intrinsic high error of long-reads technologies, Canu and Flye lead to high accurate assemblies (~99.4-99.8 % of accuracy). However, errors still had an impact on the prediction of biosynthetic gene clusters.ConclusionsMinION metagenomic sequencing data proved sufficient for assembling low-complex microbial communities, leading to the recovery of highly complete and contiguous individual genomes. This work is the first systematic evaluation of the performance of different assembly tools on MinION data, and may help other researchers willing to use this technology to choose the most appropriate software depending on their goals. Future work is still needed in order to assess the performance of Oxford Nanopore MinION data on more complex microbiomes.

No abstract

No abstract

AbstractNatural products have traditionally been discovered through the screening of culturable microbial isolates from all sort of environments. The sequencing revolution allowed the identification of dozens of biosynthetic gene clusters (BGCs) within single bacterial genomes, either from cultured or uncultured strains. However, we are still far from fully exploiting the microbial reservoir, as most of the species are non-model organisms with complex regulatory systems and yet recalcitrant to be engineered. Today, genomic and metagenomic data produced by laboratories worldwide covering the most different natural and artificial environments on Earth, are an invaluable source of raw information from which natural product biosynthesis can be accessed. In the present work, we describe the environmental distribution and evolution of the abyssomicin BGC through the analysis of publicly available genomic and metagenomic data. Our results demonstrate that the selection of a pathway-specific enzyme to direct the genome mining is an excellent strategy that led to the identification of 74 new Diels-Alderase homologs and unveiled a surprising prevalence of the abyssomicin BGC within terrestrial habitats, mainly soil and plant-associated, where we have identified five complete and 12 partial new abyssomicin BGCs and 23 new potential abyssomicin BGCs. Our results strongly support the potential of genome and metagenome mining as a key preliminary tool to inform bioprospecting strategies aiming at the identification of new bioactive compounds such as -but not restricted to-abyssomicins.