Figure 1.

Steps in the Atlas Assembly System. (1) Trim off vector and low-quality portions of reads. (2) Count k-mers in WGS reads, saving the overall distribution plus specific counts for k-mers with copy number above a threshold. (3) Align BAC and WGS reads sharing rare k-mers and save overlap edges for high-quality end-to-end alignments. (4) Enrich each BAC read set with overlapping WGS reads and their mates, assemble using Phrap, scaffold and check for consistent assembly. (5) Arrange BACs into contiguous sets (bactigs) and flag problem BACs for closer quality checking. (6) Assemble bactigs in waves designed to limit the number of BACs that are input to Phrap. (7) Treating each bactig scaffold as a unit, rescaffold to produce superbactigs, detecting problem joins and missed merges between bactigs. (8) Link superbactigs together into ultrabactigs based on remaining (single) mate-pair links, fingerprint contigs, markers and synteny with human and mouse genomes. (9) Format chromosome files with contigs separated by strings of Ns representing gaps. Quality-control feedback steps include (10) examining coassembly scores of problem BACs and removing foreign trays of reads; (11) resolving superbactig conflicts by modifying bactigs and possibly flagging BACs for closer checking; and (12) resolving ultrabactig and mapping conflicts in collaboration with research groups that generated FPC and marker information.