Impact of average edge length of de Bruijn graph on sequence correctness of assembly. A plot showing the sequence correctness of 369 bacterial genome assemblies versus the average edge length of their starting de Bruijn graphs, under three different optical map error rates. Genomes with average edge length greater than 10 kbp are generally assembled with near perfect correctness over all three error rates, while the results are mixed for genomes with shorter average edge lengths. For genomes with average edge length below 10 kbp, correctness may improve by as much as 40% when moving from the high error to low error setting, highlighting the potential benefits of more accurate mapping technologies.
Lin et al. BMC Bioinformatics 2012 13:189 doi:10.1186/1471-2105-13-189