Comparison of assembled contigs (y-axis) with the Sanger reference sequence (x-axis). Tuple plots for BACs 184G09 (A, C) and 711N16 (B, D). A and B: Mira assemblies with default parameters (hss 4, bph 16). C and D: Best MIRA assemblies after iterative parameter changes. Contigs larger than 1 kb were concatenated after directing and ordering by comparison to the Sanger reference sequence, horizontal green lines mark contig borders. Whereas in the default assembly of 184G09 (A) the longest contig is wrongly assembled in the region homolog to 102.5-108.0 kb (red rectangle) of the reference, the best assembly (C) results in a single contig accurately fitting the Sanger sequence. In contrast, the assembly problem in repetitive regions of 711N16 (B, D) even can not be resolved by the best assembly although turning out in less contigs ("collapsed" sequences, see text).