Debugging long-read genome and metagenome assemblies using string graph analysis

National audienceThird-generation long-read sequencing technologies tackle the repeat problem in genome assembly by producing reads that are long enough to span most repeat instances. In principle one expects that with such reads most bacterial genomes will be assembled into a single contig. However in practice, some datasets fail to be perfectly assembled even with leading assemblers, and are fragmented into a handful of contigs. As a mean to investigate those cases, we consider the string graphs that are generated by assemblers during intermediate stages of the assembly process. We seek to establish a coherent framework for analyzing these graphs, in the hope that they will help us determine the biological causes that led the assembler to output shorter contigs. This poster presents some preliminary results of such an analysis