Using nanopore sequencing to get complete genomes from complex samples

The advantages of “next generation sequencing” has come at the cost of genome finishing. The dominant sequencing technology provides short reads of 150-300 bp, which has made genome assembly very difficult as the reads do not span important repeat regions. Genomes have thus been added to the databases as fragmented assemblies and not as finished contigs that resemble the chromosomes in which the DNA is organised within the cells. This is especially troublesome for genomes derived from complex metagenome sequencing. Databases with incomplete genomes can lead to false conclusions about the absence of genes and functional predictions of the organisms. Furthermore, it is common that repetitive elements and marker genes such as the 16S rRNA gene are missing completely from these genome bins. Using nanopore long reads, we demonstrate that it is possible to span these regions and make complete genomes from complex samples with the throughput of a handheld DNA sequencer that plugs into a laptop. Using the kmer sequence composition signatures we demonstrate that machine learning tools can help facilitate pre-assembly sorting of the reads from complex systems. This will ultimately facilitate high throughput generation of high quality reference genomes. This is an essential step for further development of online DNA monitoring and surveillance.