Fig. 6

X-Mapper exhibits the lowest alignment inconsistency (fraction of reads reported with inconsistent alignments) and the fewest suboptimal alignments (fraction of reads reported with suboptimal alignments) when aligned to a complex reference. Alignment consistency was tested by comparing the alignment results of a whole genome sequencing (WGS) sample aligning to 88 bacterial genomes (representing a complex reference), to the alignment results of this WGS sample aligning to its own assembly (representing a simple reference). An alignment was considered consistent if: (1) the optimal alignments to the simple reference were also reported when aligned to the complex reference, or (2) the complex reference yielded better alignments (with lower penalties or higher alignment scores) than those in the simple reference. Bowtie2, Minimap2, and BWA were run under their default settings and settings that reports multiple, if not all, alignments (labeled as “all”). Since the “all” settings often report suboptimal alignments, the fraction of reads reported with suboptimal alignments was also evaluated for each aligner. Alignment consistency was tested using four different penalty settings, including the default settings of (1) Bowtie2, (2) BWA, (3) X-Mapper, and (4) Minimap2. The Bowtie2 configuration with seed mismatches was tested in this analysis