Joyjit Daw, a Principal Machine Learning Engineer at Oxford Nanopore Technologies (ONT), provided a Dorado update at the Nanopore Community Meeting Houston meeting. The Dorado v0.5.0 release supports core Guppy features, custom barcodes, read splitting, and Minimap2 alignment. Dorado has been integrated into MinKNOW. They have also extended features such as hemi methylation for the Dorado duplex. RNA methylation calling, adapter trimming, and poly(A) estimation have been implemented in a previous release. The new Dorado version has an improved simplex model. The SUP model now has a 15% reduction in error rate. There is improvement in low-complexity regions and telomere regions. There is a more accurate calling of bacterial isolates with consensus accuracy improvements. Dorado v0.5.0 has a 20% higher throughput. Daw noted that they will continue improving performance and extend GPU support. Fast5 is being replaced by Pod5 file format. I am interested in the new updates to Dorado and integration into MinKNOW.
