Fig. 1

Strategy and workflow used for assembling the optical map-- BtOM1.0. Left side describes reference-based iterative map assembly: exhaustive pairwise alignments of the complete Rmap dataset (purple lettering) against the UMD3.1 sequence reference maps (red lettering) generated in silico produces “piles” of Rmaps. Such alignments are then divided into overlapping bins (1 Mb bins; 500 Kb overlap), which are then independently assembled into updated reference maps (optical contigs bearing consensus maps) that are used for 8 subsequent cycles (blue circular arrows) of alignment (entire Rmap dataset) and assembly, all performed without using the sequence reference. Right side depicts de novo map assembly, using those Rmaps from the entire dataset not recruited for optical map contig formation during iterative map assembly, which are used to construct a de Bruijn graph via k-mer hashing. “Seed” maps are then assembled from Rmaps in each each confident node within graph and used as an optical reference for pairwise alignments (piles) of Rmaps during iterative assembly (8 times; blue circular arrows). Bottom shows the merged assembly of updated optical contigs from each bin (iterative assembly) and contigs assembled from “seed” maps (de novo assembly) into the final optical map-- BtOM1.0, which was used to tabulate map vs. sequence discordances