strain and 15-fold coverage for the type strain. From the quality histograms, the red lines represent 2D reads' scores which are clearly better than the ones in 1D reads (green and blue for template and complement sequences respectively). The length distribution shows that MinION sequencing returned significant reads with length around 5K for the first sample and 7K for the latter. The statistics on the far right of the table show error rate in term of indels, mismatches and hit rate from alignments with reference data. Overall, chemistry R7.3 is more up-to-date than R7 and as expected, gave slightly better results. As a next step, there came a mission to assembly those reads, together with available Illumina data of corresponding samples, in an attempt to have complete genome reconstructions for these two K. pneumoniae strains. This task was important at the time as it would depict bigger picture about interactions between genetic elements in the whole genome and help to comprehend the biological pathways supporting this superbug's mechanisms. 1.3.1 Definition Despite the huge improvements of sequencing technology, there still exist the common limitation that in most cases, it is impossible to sequence directly the whole length of a genome. As shown in Figure 1.5, the actual practice is to break the whole genome into smaller pieces 1.4 Genome assembly with long read data Nanopore data can be used on its own for de novo genome assembly, or in conjunction with high quality SGS data in a process called hybrid assembly. Both approaches have their own pros and cons and should be considered in reference to available dataset and the ultimate purposes of using the final assembly. This is unlike SGS sequencing-by-synthesis platforms, which sequence billions of short reads in parallel, with each cycle (contributing an extra base) taking several minutes, with the data being provided in batch at the end of a run. For this task, the raw data of a read must be retrieved and analyzed while sequencing is