Fig. 3

Scaffolding high-quality short read assemblies with Oxford Nanopore Technologies long reads. Publicly available ONT long reads for E. coli K-12 MG1655, S. Typhi and S. cerevisiae W303 were recently made available [6–8]. We have used these data to re-scaffold E. coli K-12, S. cerevisiae S288c and S. cerevisiae W303 baseline assemblies of Illumina-only data using LINKS, AHA and SSPACE-LR and assessed the quality of the resulting assemblies by plotting the NG50 length contiguity and number of misassemblies reported by QUAST [24] (black square, green triangle, blue diamond). Also, we have re-scaffolded a baseline S. Typhi Illumina assembly iteratively (11x) with LINKS using R7 2D ONT data and compared it to the SPAdes hybrid Illumina + ONT assembly reported [7] (red circles). We compare the results to Celera Assemblies (CA) of Illumina-corrected ONT reads (NaS and Nanocorr polished) and ONT-corrected ONT reads (Nanocorrect/Nanopolish) [4, 8, 26]. The re-scaffolding software ran on either all R7 chemistry 2D (2D), raw or Nanocorr-corrected reads, as indicated. For S. Typhi, the AHA and SSPACE-LR re-scaffolded assemblies were comparable and their corresponding data points overlapped (refer to Table 2). Data point size is normalized on the number of resulting scaffolds in each experiment. Smaller points indicate a better outcome (less scaffolds)