Ancora: a web resource for exploring highly conserved noncoding elements and their association with developmental regulatory genes

Abstract

Metazoan genomes contain arrays of highly conserved noncoding elements (HCNEs) that span developmental regulatory genes and define regulatory domains. We describe Ancora http://ancora.genereg.net, a web resource that provides data and tools for exploring genomic organization of HCNEs for multiple genomes. Ancora includes a genome browser that shows HCNE locations and features novel HCNE density plots as a powerful tool to discover developmental regulatory genes and distinguish their regulatory elements and domains.

Figure 1

Comparisons available in Ancora. Shaded boxes correspond to genomes shown in the Ancora genome browser. Connecting lines indicate pairwise genome comparisons for which HCNEs are available in Ancora. The following genome assemblies underlie the current data sets: human NCBI 36, mouse NCBI 36 and 37, chicken v2.1 [41], Xenopus tropicalis v4.1 (US DoE Joint Genome Institute), zebrafish Zv6 and Zv7 (The Wellcome Trust Sanger Institute), fugu v4.0 [42], Tetraodon nigroviridis V7 [19], stickleback v1.0 (The Broad Institute), medaka v1.0 [43], D. melanogaster rel. 5 [44], D. pseudoobscura rel. 2 [45] and the February 2006 releases of D. ananassae, D. virilis and D. mojavensis [46].

Figure 2

A 1.7 Mb region around the human SHOX2 gene. (a) Ancora genome browser main view. SHOX2, a homeobox gene implicated in limb development [47], is embedded in an array of HCNEs detected by comparison with mouse and zebrafish genomes. Overlaid density plots show densities of HCNEs detected at similarity thresholds of 95% (yellow), 98% (orange) and 100% (red) in the mouse comparison and similarity thresholds of 70%, 80% and 90% in the zebrafish comparison, over a 50 column sliding window. Note that the density of the most strongly conserved HCNEs (red) peaks around SHOX2. Synteny blocks are based on net alignments with the zebrafish genome [18]; boxes indicate aligned segments, connecting lines indicate gaps and labels indicate alignment orientation and position in the zebrafish genome assembly. The centrally shown synteny block encompasses SHOX2, RSRC1 (a gene of unknown function) and the array of HCNEs conserved in zebrafish. (b) Conservation profiles for the same region in the UCSC Genome Browser [21]. Comparison between (a) and (b) demonstrates qualitatively different information provided by the HCNE density plots in (a).

Figure 3

HCNE track configuration page. Up to three HCNE sets from each pairwise comparison can be shown simultaneously. A set is selected by choosing a similarity threshold (for example, 70% identity over 50 alignment columns), and can be further restricted by an arbitrary threshold on HCNE size. Note that HCNEs may be larger than the window size (30 or 50 columns) used to identify them because the procedure that detects HCNEs merges overlapping conserved elements. For each selected set, the user can choose to see HCNE densities, HCNE locations, or both. Density plots for the different sets will be overlaid (Figure 2a), so that the plot for set two is drawn on top of that for set one, and the plot for set three drawn on top of that for set two. If the option to separate densities based on chromosomes in other genomes is used, the browser will attempt to create one density plot for each chromosome (in the other genome) for which there are HCNEs in the displayed region, or within half a sliding window to either side. If the resulting number of plots exceeds the number of plots requested on this configuration page, densities for the chromosomes with least HCNE sequence in this region will be combined into one plot labeled 'other' (Figure 5).

Figure 4

HCNE density distributions on human chromosome 3. Shown are densities of HCNEs identified from comparison with mouse, chicken and three different fish genomes. This genome browser screenshot has been manually labeled with likely target genes of HCNE enhancer activity at major density peaks. Target genes were identified by zooming in to inspect gene annotations at each peak.

Figure 5

Duplicated GRBs. Zebrafish HCNEs and their density distribution are shown for the human (a) PAX7 and (b) LHX1 loci. HCNEs are colored by the zebrafish chromosome they map to. In (a), most HCNEs are colored light green or gray and map to zebrafish chromosomes 11 or 23, respectively. In (b), most HCNEs are colored blue and map to zebrafish chromosome 15 (because this region contains many HCNEs, they are collapsed on a single row in this screenshot). The density plots are also separated based on zebrafish chromosomes. Comparison of synteny blocks to exon locations indicate that orthologs of AATF and ACACA are present next to the LHX1 ortholog (lhx1b) on zebrafish chromosome 5, but not on zebrafish chromosome 15 where lhx1a is located; this can be confirmed by detailed inspection of the zebrafish loci.

Figure 6

Distinguishing regulatory domains. Screenshot from the Ancora genome browser showing 26 Mb of human chromosome 10 and HCNE densities from comparisons with mouse, zebrafish and medaka. HCNE density curves are separated based on chromosomes in these organisms (Zv7_NA53 is a contig that has not been assigned to a chromosome). To illustrate the use of this view for distinguishing chromosomal regulatory domains, rectangles have been manually added to the screenshot around density peaks indicating clusters of HCNEs in conserved synteny. Rectangles are labeled with regulatory genes annotated in the corresponding genomic regions.

Figure 7

Ancora tracks in the UCSC and Ensembl genome browsers. Genome browser views of region around the human SHOX2 gene. Added tracks show locations and densities for HCNEs detected at similarity thresholds of 95% in the mouse comparison and 70% in the zebrafish comparison. (a) UCSC, same region as in Figure 2. (b) Ensembl, displaying 1 Mb (the maximum allowed size in ContigView) of the same region. Additional tracks from Ensembl show conserved elements ('Conservation') and transcripts ('Ensembl trans.').