|Home | About | Journals | Submit | Contact Us | Français|
Summary: Comparative genomics remains a pivotal strategy to study the evolution of gene organization, and this primacy is reinforced by the growing number of full genome sequences available in public repositories. Despite this growth, bioinformatic tools available to visualize and compare genomes and to infer evolutionary events remain restricted to two or three genomes at a time, thus limiting the breadth and the nature of the question that can be investigated. Here we present Genomicus, a new synteny browser that can represent and compare unlimited numbers of genomes in a broad phylogenetic view. In addition, Genomicus includes reconstructed ancestral gene organization, thus greatly facilitating the interpretation of the data.
From less than 20 fully sequenced metazoan genomes 4 years ago, nearly 80 species are now represented in a variety of centralized databases. This abundance of sequence data has reinforced the role of comparative genomics as the primary approach to gain insight in the organization of a genome. Comparing sequences from different species serves several purposes: (i) to outline conserved regions, a powerful guide to rapidly focus on functional regions; (ii) to document differences among these functional sequences as a first step to understand broader biological differences (metabolic, developmental, etc.) between organisms; and (iii) to identify evolutionary events that have interrupted the gene colinearity between the genomes of two species since their last common ancestor.
To document and study the latter, the inference of ancestral gene orders starting from extant species provides important reference points; yet no visualization tool currently allows comparisons between an ancestral genome to one or more of its modern descendant. Existing software still limit the comparison to two or three extant genomes at a time, and are restricted to a limited range of species (Byrne and Wolfe, 2005; Courcelle et al., 2008; Derrien et al., 2007; Dong et al., 2009; Jensen et al., 2009; Lyons et al., 2008; Pan et al., 2005; Sinha and Meller, 2007).
To address these issues, we have developed Genomicus, a browser dedicated to the study of synteny and the conservation of gene order among multiple genomes (currently 52 metazoan genomes and the yeast Saccharomyces cerevisiae). Importantly, Genomicus also integrates reconstructed ancestral synteny blocks at 44 ancestral nodes.
Most of the genome data displayed in Genomicus is already stored, integrated and publicly available from the Ensembl database (Hubbard et al., 2009) but without extensive synteny visualization tools. The two main types of information that are required by Genomicus are gene positional information in their respective genomes and phylogenetic relationships (orthology, paralogy) between genes. Genomicus then edits Ensembl phylogenetic trees (Vilella et al., 2009) in three ways. First, duplication nodes with a Duplication Consistency Score (Vilella et al., 2009) below a threshold, that is optimized to increase the synteny between extant genomes, are selected. In such cases, duplication nodes are shifted towards terminal branches unless stopped by an intermediate, strong, duplication node. Second, we have added Boreoeutheria, Euarchontoglires and Atlantogenata ancestral nodes in existing trees of placental mammals (Prasad et al., 2008). Third, we have added some extant species that are not currently referenced in Ensembl (Branchiostoma floridae, Nematostella vectensis and Oikopleura dioica), together with their respective ancestral nodes. For each of these new species, best reciprocal blast comparisons [best reciprocal hit (BRH)] are performed between predicted proteins and the proteins from a set of key species already referenced in Genomicus. Comparisons that are internally consistent (mutual orthology relationships are respected) allow a given protein to be added in the same phylogenetic tree as that of its BRH. In rare cases, a new protein may act as outgroup to two existing trees and fuse them through a new duplication node.
Ancestral syntenic blocks are reconstructed by a complex procedure that will be described in details elsewhere (M.Muffato et al., manuscript in preparation). Briefly, parsimonious scenarios are estimated based on pairwise comparisons of gene order between all available sequenced genomes (1378 comparisons in Genomicus v56.01). For a given ancestor, all ancestral genes that are identified as conserved neighbours in at least one such comparison become linked nodes in a graph. A weight (with values comprised between 1 and 1378) reflecting the number of times this situation was observed in all the comparisons is then applied to each link.
At this stage, inconsistencies may appear in the form of ancestral genes connected to more than two neighbours. To resolve these, the weighted graph is processed using a top-down greedy algorithm where the links of highest weight are selected first and are used to select the most likely gene-to-gene connection in case of multiple choice.
This produces a set of linear paths in the graph connecting ancestral genes based on the number of times their respective descendants are observed as extant neighbours. We performed extensive simulations and benchmarked our methods against several alternative methods: MGR (Bourque and Pevzner, 2002), MGRA (Alekseyev and Pevzner, 2009) and InferCars (Ma et al., 2006). Our method is the only approach able to satisfactorily analyse data with the volume (53 species and 888 217 extant genes) and complexity (duplications, deletions) found in the complete set of sequenced vertebrate genomes. The reconstructed gene order is correct in >95% of the cases (specificity), and includes between 70% and 95% of the expected ancestral gene pairs (sensitivity).
The home page invites the user to enter its gene of interest and will by default show a graphical representation in PhyloView.
In both views, the tree can be edited (by expanding, collapsing, hiding, showing chosen nodes) to clarify the view. Genomicus also displays orthologous conserved non-coding elements (CNEs) at three levels of conservation. Finally, gene and loci information can be reached with links to other browsers such as Ensembl, UCSC and NCBI.
The main perspectives are to extend the functionalities and the breadth of species displayed in Genomicus. In particular, a ‘chromosome painting’ view showing extant and ancestral karyotypes that are colour coded according to a species of interest is currently in development. Genomicus will also follow the ‘Ensembl Genomes’ project and will therefore extend its scope to include plant and fungal genomes.
We thank Pierre Vincens and his group at the Ecole Normale Supérieure for providing hosting and security services for this project.
Funding: Agence Nationale pour la Recherche (ANR-07-GANI-008-01).
Conflict of Interest: none declared.