The Genom of Homo sapiens.pdf

Evolving Methods for the Assembly of Large GenomesR.A. GIBBS AND G.M. WEINSTOCKHuman Genome Sequencing Center, Baylor College of Medicine, Houston, Texas 77030Several large genomes have been sequenced in recentyears, leading to the general perception that genome assemblyis a solved problem, and, aside from the considerableexpense of generating the DNA sequence reads,the pathway to completing future projects is straightforward.Closer examination reveals that this is not the case,and the complexity of large genomes and the relativenewness of the tools available for piecing them together,plus the rapid evolution of related technologies, make thisan active area for scientific development. A particularconcern is the ongoing role of mapping and utilization oflarge insert bacterial artificial chromosome clones(BACs) when tackling new species for which resourcesand reagents are scarce.Current approaches for assembling large genomes includethe “hierarchical” strategy, used for the publiclyfunded nematode consortium (The C. elegans SequencingConsortium 1998) and human genome projects (Landeret al. 2001), and the whole-genome shotgun (WGS)methods (Weber and Myers 1997) used for the privatelyfunded human sequence (Venter et al. 2001), Drosophilamelanogaster (Adams et al. 2000), mouse (Waterston etal. 2002b), and, more recently, ciona (Dehal et al. 2002),zebrafish (http://www.sanger.ac.uk/Projects/D_rerio/),Drosophila pseudoobscura (http://www.hgsc.bcm.tmc.edu/projects/drosophila/), the chimp (Pennisi 2003),pufferfish (Aparicio et al. 2002), dog (Kirkness et al.2003), and honeybee (http://www.hgsc.bcm.tmc.edu/projects/honeybee/). Conceptually, these methods representopposite extremes (Fig. 1). In the hierarchical strategy,considerable effort is required at the beginning of theproject and generally includes the generation, manipulation,and analysis of BACs. Ideally, the precise positionalrelationship between the BACs is established prior to sequencing.Next, each individual ~200-kb clone is treatedBAC by BACWhole GenomeShotgun (WGS)as an individual, localized DNA sequence project. Thisincludes random sequencing as well as a more painstakingfinishing phase. Finally, the BAC sequences arejoined at the end of the project, to reconstruct the completegenome.In contrast, in a WGS project, sequences are generatedfrom the ends of inserts of randomly selected (shotgunned)subclones of the whole genome. When a sufficientaverage depth of sequence coverage is achieved, thegenome is assembled by advanced software that primarilyfunctions by recognizing individual sequence overlaps.To maintain assembly accuracy over long genomedistances, the sequences are generated from subcloneswith different insert sizes. The assembly is subsequentlyconstrained by requiring that sequence reads from theends of the same subclone (mate-pairs) are at their expectedrelative positions in the final sequence (Edwardset al. 1990). In many instances, these clone-end sequencesinclude at least a small fraction of BAC ends,since those clones are easily manipulated and their insertsspan hundreds of kilobases.Depending on the depth of coverage, a randomly sequencedgenome can be referred to as a preliminary draft(