Katherine Lawrence, from the Machine Learning and Bioinformatics team at Oxford Nanopore Technologies, spoke at the Nanopore Community Meeting on “Bacterial isolate and plasmid sequencing.” Lawrence began by emphasizing the importance of microbes and learning about their sequences. Long reads help assemble bacterial genomes efficiently. Bacterial genome modifications include 4mC, 5mC, and 6mA. Lawrence noted that methyltransferases in bacteria bind to specific motifs and can be transferred between organisms. Lawrence went on to explain Nanopore signal signatures of modified bases. Training basecaller models to detect these modifications helps improve accuracy. With data from the Nanopore Community, improved models have been created. Consensus polishing steps with Medaka further improve strand-bias errors. A Medaka model for bacterial methylation provides additional improvement of the consensus sequence. This polishing model also improves errors in plasmids. Medaka updates are included in EPI2ME workflows and the stand-alone version. I want to take advantage of these updates and learn more about the genomes of Delftia isolates!
