Automated strain separation in low-complexity metagenomes using long reads

Starting on
Ending on
IRISA Rennes
Riccardo Vicedomini (Institut Pasteur)

Recent methodological and technological advances enabled the reconstruction of bacterial genomes from complex microbial communities and, to a certain degree, a strain-level characterization. Nevertheless, at present, methods aiming to characterize metagenomes at the strain level are based on either short-read data or hybrid approaches. This motivated us to develop Strainberry, an assembly-based method that separates individual strains in low-complexity metagenomes using uniquely long reads. We benchmarked Strainberry on mock communities and real datasets. It provided strain-resolved assemblies in low-complexity metagenomes, but was also able to unravel a more fine-grained microbial diversity in samples of higher complexity.

For the Symbiose groups.