The Genom of Homo sapiens.pdf

DETECTING MULTISPECIES CONSERVED SEQUENCES 257pq22.322.121.321.115.315.114.314.11312.312.111.211.111.2111.2211.2321.1121.1321.322.122.231.131.3131.3332.233343536.136.236.37Figure 2. Features of two targeted genomic regionssequenced in multiple species. The two indicatedregions of human chromosome 7 weresubjected to multispecies comparative sequencing(see http://www.nisc.nih.gov for details). Thegeneral features of each region are listed, with theindicated numbers corresponding to the humanreference sequence. “No. Known Genes” is thenumber of annotated genes with defined codingsequences in each region. “Total Mb Sequenced”reflects the amount of sequence data generated inaggregate from the indicated number of differentspecies. “Percent GC” is the average GC content(calculated in nonoverlapping 1-kb windowsacross each region). Also indicated is informationabout the MCSs detected in each region (seetext). For the 7q31.3 region, the MCSs were identifiedusing the sequences from a subset of 13species (among those excluded were all primates,tetraodon, and zebrafish); for the 7q11.23 region,sequences from all 12 species were used to detectMCSs. The last four rows reflect the percent ofMCS bases that overlap coding, UTRs, ARs, andnoncoding sequence, respectively.Detected MCSs General Features7q11.23 7q31.3Size (Mb) 6.3 1.8No. Known Genes 42 10No. Species Sequenced 12 25Total Mb Sequenced 29 31Percent Coding 0.9 1.2Percent UTR 0.5 0.6Percent Repeats 55.1 40.3Percent GC 49.1 38.4No. 4,520 1,572Average Length (bp) 70 60Percent of Total Sequence 5 5Percent Coding 16 22Percent UTR 3.4 4Percent AR 3.6 3.6Percent Non-Coding 77 70represents the most detailed comparison of vertebrategenomes performed to date. Several findings from this effortare particularly relevant to our multispecies sequencingprogram. First, roughly 40% of the mouse genomeforms sequence alignments with the human genome usingestablished alignment methods (e.g., blastz [Schwartzet al. 2003a]; see the human–mouse alignments depictedin Fig. 3). Second, only about 5% of the mammaliangenome is estimated to be actively conserved (Waterstonet al. 2002; Roskin et al. 2003), and this consists ofroughly 1.5% that is protein coding and 3.5% that is noncoding.At present, the specific bases that constitute this5% are not known. Thus, a critical challenge is to developstrategies for identifying this small fraction of activelyconserved sequence, since it presumably reflects the bulkof the functionally important portion of the humangenome. However, simple pair-wise comparisons of humanand mouse sequences are ineffective at identifyingthe correct subset of actively conserved sequence in themammalian genome (Thomas et al. 2003).To broaden the above human–mouse sequence comparisons,we have performed similar pair-wise alignmentsbetween human and various other species’ sequences(Fig. 3), revealing a number of interestingfindings. First, primate sequences are highly similar tothe human sequence, as expected. Second, alignments betweenthe human sequence and that of the other mammalsshow patterns similar to that seen with mouse; however,the sequence conservation with human is generallyhigher for most nonrodent placental mammals comparedto rodents (Thomas et al. 2003). Alignments between humanand fish sequences are almost exclusively confinedto coding regions. Interestingly, marsupial, monotreme,and chicken sequences show significantly fewer alignchromosome7q31.3 (Thomas et al. 2003) and 7q11.23, respectively(Fig. 2; also see http://www.nisc.nih.gov/data).Together these regions span >8 Mb in the human genomeand contain 52 known genes; they are also notably differentwith respect to their general sequence composition(e.g., GC content, amount of repetitive sequences, and proportionof coding sequence).The sequences and associated analyses emanating fromthe NISC Comparative Sequencing Program reflect increasinglycomplex data sets, especially with respect tothe number of vertebrate species represented and the variouscomparisons being performed. For visualizing anddisseminating these data, we have collaborated with ourcolleagues at the University of California, Santa Cruz(UCSC) to facilitate their development of a “zoobrowser” component of the UCSC Genome Browser(Kent et al. 2002; Karolchik et al. 2003; Thomas et al.2003; also see http://genome.ucsc.edu). A sample displayof this browser for a small portion of the 7q31.3 target isshown in Figure 3. In addition to providing convenientaccess to the assembled sequences for each species, thebrowser has been customized to display the results ofpair-wise sequence alignments (which can be performedusing any species’ sequence as the reference) and othercomparative analyses (see below). By direct integrationwith the standard UCSC Genome Browser environment,such data can then be viewed within the context of theever-growing collection of annotations and associated informationabout the human genome sequence.MULTISPECIES CONSERVED SEQUENCESThe recent comparative analysis between the humanand mouse genome sequences (Waterston et al. 2002)