Figure 3

Details of the Barnacle pipeline. A) Contrasting a collinear alignment topology (i) with non-collinear topologies: (ii) interchromosomal, which involves alignment to two chromosomes; (iii) inversion, which involves alignment to two strands; (iv) eversion, which involves alignment with a reversal of block ordering; and (v) duplication, which involves multiple alignment to the same region. B) (i) Pieces of the contig can be aligned to different regions in the genome, with ‘q’ denoting the quality of each alignment, normalized to the range [0,1]. (ii) Alignments 1 and 5 are selected, because of their high qualities and inclusion, and their low overlap. C) Alignment selection can result in one of four cases: (i) a single ungapped alignment is selected, (ii) a single gapped alignment is selected, (iii) a pair of alignments is selected, or (v) more than two alignments are selected. D) In gap contigs a piece of the contig does not take part in the initial contig-to-genome alignment. Gap contigs are checked for duplications (i) by realigning the gap sequence back to the contig with the original gap location masked, and for inversions (ii) by realigning the gap sequence to a region of the genome determined by the original contig-to-genome alignment. E) Fusions can have homologous sequence near the breakpoint that makes it impossible to determine the precise breakpoint position. F) For split candidates (i), read support is calculated in the region surrounding the overlap of the two contig-to-genome alignments. For gap candidates involving a duplication (ii), read support is calculated in the region between the two copies of the duplicated sequence.