Fig. 1

Comparison of chloroplast genome assemblies. The coverage of the long- and short-reads is shown along the top and left-hand-side of each panel, respectively. The left-hand-side also shows which assembler was used for each row of assemblies in that panel. Hinge and Canu are long-read-only assemblers, whereas Unicycler is short-read-only and hybrid assembler. The Hinge and Canu results in b, c, d were polished by Racon+Nanopolish, and the Unicycler results used Karect-corrected short-reads. a. The total coverage of the chloroplast genome across all contigs output by the assembler. Panels marked with a red ‘x’ contained a single contig covering the whole chloroplast genome. The heatmap indicates the chloroplast genome coverage. b. The assembly length of different assemblies after manual curation (e.g. removing duplicate regions). Panels marked with an ‘x’ denote assemblies with the expected length, in the range 155,938 bp–155,945 bp. c. The mapping rate of validation reads to the assemblies after manual curation. Assemblies with highest mapping rate (99.43%) are marked with a red ‘x’. d. The average per-base error rate of validation reads mapped to each manually-curated genome assembly. Assemblies with the lowest error rate (0.0007) are marked with a red ‘x’