Given the large quantity of data incorporated, the first-generation integrated and virtual genome maps reported here will enhance significantly genome research in the tammar wallaby (a valuable model kangaroo species), and facilitate the assembly of the genome sequence of this species.
Whenever comparative data were required, we have been conservative in using wherever possible the genome of the most closely related sequenced marsupial. The tammar wallaby and the opossum diverged around 70 mya , comparable with the divergence within eutheria, amongst which much use has been made of comparative information, e.g. dog and human ; sheep and human . The next-best choices are eutherians, which are more than twice as distant (diverging 150 mya) [9, 10]. These realities provide a strong justification for the present strategy of drawing comparative information from opossum in preference to eutherians.
In addition, the remarkable conservation of chromosome arrangement in marsupials makes this approach particularly appropriate for the tammar wallaby. Whereas the eutherian genome has been grossly rearranged in many lineages, there is very strong conservation of synteny between tammar wallaby and opossum , even to the extent of whole chromosome arms being conserved . When the integrated map was used to create the virtual genome map, the preferred strategy was, once again, to rely as much as possible on the tammar wallaby's nearest sequenced evolutionary neighbour, and then to turn to one of the most mature genome assemblies, namely human, only in the minority of cases where the opossum information was not sufficient.
Of course there will be errors in the order and relative location of loci, especially in the virtual genome map: the integration of data from conserved blocks of synteny means that the location of most tammar wallaby genes in the virtual genome map is predicted on the basis of their order in other species. However, as argued above, in using whenever possible comparative mapping data from opossum, we are, in effect, relying on better comparative information then has been the basis of the utilisation of comparative mapping information within eutherians.
Estimation of conserved-synteny blocks is not a simple process, and errors are certain to have been made, given the relative paucity of information available in the tammar wallaby. For the markers with the lowest confidence level, it is important to note that close-range locus order presented is just one of several equally likely possibilities.
As more sequence-level comparative data become available, these blocks will be better defined.
In principle, the overall aim of creating an integrated map is to combine together in a rational manner all available mapping information in the species of interest, without recourse to any information from other species. In the creation of the integrated map of sheep, for example , comparative data were used only in the local repositioning of loci that had all been FISH-mapped to the same chromosomal band. As discussed by Liao et al. , this did not compromise the essential integrity of the integrated map in reflecting all available sheep data: it simply provided a first estimate of the order of a set of loci that are known to be located within a particular band. In the case of the tammar wallaby, there was a lack of orthologues mapped with sufficient resolution in this species, which precluded the local ordering of loci that had been FISH-mapped to a particular band. The best solution, given the lack of resources to create a denser physical map, was to estimate evolutionary breakpoints in the tammar wallaby with respect to the opossum (wherever possible) and human assemblies. Whilst this provides an additional compromise to the integrity of the integrated map, this does not alter any mapping data gleaned from the wallaby alone. Therefore, in practice, it does not provide any additional compromise to the integrity of the wallaby-mapping data.
How does this first-generation virtual genome map compare with the resources used in genome assemblies in other species? The bovine genome sequence  was assembled onto a single RH map . The opossum genome assembly  was assigned to chromosomes based primarily on FISH-mapping of BACs from scaffolds , with support from the second of two linkage maps whose terminal markers had also been FISH-mapped . Our tammar wallaby first-generation virtual genome map is more comprehensive than either of these strategies, since it is based on all available mapping information from the species itself, combined in a rational manner, supplemented by comparative mapping data. This integrated map is better and more useful than either of its components considered alone. Obviously it would be desirable to obtain more mapping information (both linkage and physical) for the tammar wallaby. As such data become available in the future, they will be used in the construction of second-generation integrated and virtual genome maps. In the meantime, the maps described in this paper are the best available at this time; they utilize all available information to create the most complete maps of the tammar wallaby chromosomes that can be produced at this time.
As pointed out by Lewin et al. , "Every genome sequence needs a good map". Genome sequence itself is not sufficient to enable a chromosome assembly or construction of good comparative maps to reveal hidden evolutionary stories. Good genome maps (e.g. physical maps, RH maps, linkage maps) are a necessary complement to genome sequence. However, they are of limited use in isolation. What is needed is a means of integrating all available mapping data for a species into a single map. The first-generation integrated map reported in this paper achieves this aim for the tammar wallaby, and has enabled the creation of a first-generation virtual genome map for this species, combining the integrated map with comparative mapping data from species with more mature chromosome assemblies.
By combining the first-generation virtual genome map presented in this paper with the Ensembl annotation  of the initial (Meug_1.0) tammar wallaby assembly, it is now possible to construct the first draft chromosome assembly for the tammar wallaby. In their annotation process, Ensembl were able to create 10257 "gene-scaffolds" comprising two or more Meug_1.0 scaffolds. Of these, 7027 have one gene in common with the virtual genome map, and an additional 953 have more than one gene in common with the virtual map, giving a total of 7980 gene-scaffolds that can be incorporated into a chromosome assembly, based on the virtual genome map. Thus 78% of the Ensembl gene-scaffolds can be incorporated into a tammar wallaby chromosome assembly, and 9% of the gene-scaffolds can be orientated in this chromosome assembly.
In addition to the Ensembl gene-scaffolds, there are another 1175 Meug_1.0 scaffolds that have at least one gene in common with the virtual genome map, and 54 Meug_1.0 scaffolds that have multiple genes in common with the virtual genome map. The total size of gene-scaffolds and scaffolds that can be incorporated into a chromosome assembly is 533,684,520 bp, which is 22% of the estimated tammar wallaby genome size (2457 Mb). This chromosome assembly includes 10522 of the 15290 protein-coding genes identified in the Ensembl annotation. In other words, the virtual genome map enables the creation of a chromosome-based tammar wallaby genome assembly that includes a high proportion (69%) of protein-coding genes identified in the sequence data. This compares with the few gene-scaffolds whose location can be determined solely from the integrated map built almost exclusively from tammar wallaby mapping information: only 265 gene-scaffolds have one gene in common with the integrated map and three gene-scaffolds have more than one gene in common with the integrated map. Also the virtual genome map has been tested in the recent tammar wallaby genome sequence assembly attempt and has significantly enhanced the N50 of the assembly .