Rapid development of genomic resources in fish species has provided the opportunity for comparative genome analysis, shedding lights on the structure, organization, function and evolution of vertebrate genomes. In this study, we conducted the whole genome comparative analysis of channel catfish, an important aquaculture species, with several model fish species. By comparing with other fully sequenced model fish species including zebrafish, medaka, stickleback, and Tetraodon, homologous chromosomes among these species were determined and a large number of conserved syntenies were identified, providing valuable information for whole genome assembly and annotation in catfish, and for comparative genome analysis of other teleost species.
Comparative map is a powerful tool in genomics studies, especially for non-model species, by transferring the genomic information from well-studied model species. Comparative map not only allows better understanding genome arrangement during evolution, but also benefits the discovery or confirmation of orthologies among species.
Without a well assembled whole genome sequences, comparative genome analysis can be achieved by using various other genomic resources containing information for genome level conservation. Markers with low levels of conservation in evolution have very limited value for comparative genome analysis, and most often only for very closely related species. For instance, when microsatellite markers on linkage maps were used for comparative analysis, a small number of microsatellites could be successfully mapped, indicating relative low levels of conservation of microsatellites which were derived from non-coding regions of the genome . Comparative analysis among different species using gene-derived markers on linkage maps was more effective because genes are well conserved through evolution . Higher resolution of comparative maps can be achieved by using integrated physical and genetic linkage maps with BAC end sequences. For instance, Zhang et al.  conducted comparative analysis of one catfish chromosome (linkage group 8) with four model fish species utilizing catfish linkage map, physical map, BAC end sequences and draft genome sequences. In that work, 287 unique genes were identified, and a number of conserved syntenies were identified. Although that work demonstrated the utilities of linkage maps when integrated with physical maps with BAC end sequences, the ability to establish whole genome comparative map was hindered by the lack of internal BAC sequences. In order to increase the power of comparative genome analysis, we recently developed one additional valuable genome resource, the physical map contig-specific sequences . In this study, we used all the existing genome resources of catfish for whole genome comparative analysis. The ability to identify long conserved syntenic blocks was much enhanced. For instance, with the same linkage group, LG8, as used in Zhang et al. , we were able to identify 585 unique genes, more than the double of the 287 genes identified in the previous study , and the size of syntenic blocks were much increased.
The strategy developed in this study allowed whole genome comparative analysis of catfish conducted without a well-assembled whole genome sequence. However, lacking of a continuous reference sequences, the order and orientation of catfish draft genome contigs/scaffolds within the same catfish physical contig cannot be determined at present. The reason for this inability is the low resolution of the linkage map. Because of the low resolution of the linkage map, many genes are mapped directly or indirectly to the same map location, forming stacks of sequence contigs and scaffolds with gene orders and orientations undetermined. Therefore, it is imperative to develop linkage maps with many more markers or using high density marker such as SNPs in the future, and more importantly with high resolution by using large resource families.
Comparative analysis was conducted based on the similarities between gene sequences in catfish and the homologous genes in the genomes of other model fish species. Only genes were used for analysis in this study because gene sequences are more conserved than intergenic sequences. The largest number of homologous genes was found in zebrafish among four model fish species, with 14,035 genes, followed by stickleback with 10,430 genes, medaka with 9,949 genes and Tetraodon with 9,920 genes. This difference of the homologous genes identified in different model fish species may be resulted from two reasons: First, the quality of the reference genome. For instance, of all annotated 32,574 zebrafish genes, only 1,225 (3.7%) are unmapped onto zebrafish chromosomes or mitochrondria, while in Tetraodon, there are 6,487 (31.5%) of all 20,562 genes cannot be mapped (Ensembl database). Second, the phylogenetic relationship between catfish and these model fish species determined that zebrafish is the most closely species to catfish [63, 69–71].
High levels of chromosomal conservations were observed between catfish and the four fish species. However, due to the difference of chromosome numbers among those fish species, e.g. catfish has 29 chromosomes, while zebrafish has 25 chromosomes, Tetraodon has 21 chromosomes, medaka has 24 chromosomes and stickleback has 21 chromosomes, chromosome breakage or fusion would have occurred during evolution. For instance, 33% of genes identified on catfish LG19 were found to be homologous in zebrafish chr.16, and 46% of genes were found to be homologous in zebrafish chr.24 (Figure 2), suggesting catfish LG19 have been created by fusion of chromosomal segments similar to zebrafish chr.16 and chr.24, or inversely the two zebrafish chromosomes have been created by split of the chromosome similar to catfish LG19. Similar cases can be observed between the comparison of catfish and medaka, catfish and Tetraodon, catfish and stickleback, indicated that chromosomal fusions or splits occurred frequently during the teleost evolution.
Sarropoulou et al.  conducted a comparative study in which the syntenic relationship between six non-model fish species genomes were established by using ESTs and microsatellites sequences. Our study here extended that study by adopting a much larger numbers of genes. For instance, catfish LG15 was identified to be the homologous chromosome of medaka chr.7 (C7), Tetradon chr.9 (T9), and stickleback Grp. XII (SXII), respectively, which indicated that C7, T9 and SXII were homologous chromosomes to one another. This was consistent with the results of Sarropulou et al. . Similarly, catfish LG9 (C9) corresponded to Tetraodon T1, T2 and T13 in this study, while T1 was reported to be homologous chromosome to medaka M10 , which also had the highest percentage homologous gene hits to catfish LG9 in our study.
Because zebrafish is the most closely related model fish to catfish, detailed comparative analyses were conducted between them. Catfish has 29 pairs of chromosomes while zebrafish has 25 pairs of chromosomes. Therefore, some zebrafish chromosomes are expected to be homologous to greater than one chromosome in catfish. This was found with several chromosomes (Figure 2). However, the opposite situation was also found with one catfish chromosome being homologous to several zebrafish chromosomes (Figure 2), suggesting extensive chromosome rearrangements during evolution.
A large number of conserved syntenic blocks between catfish and zebrafish were established. Analysis of the conserved syntenies should greatly benefit genome annotation in catfish. This is particularly true when dealing with large gene families and duplicated genes. As reported by Liu et al. , identities of genes involved in large gene families such as the ABC transporter gene families sometimes cannot be resolved by phylogenetic analysis alone. Syntenic analysis is essential to provide orthologous information for the identification of such genes. The inferred orthologies are important not only for the identification and annotation of genes, but also for functional inference based on orthologies . Apparently, catfish genome is well conserved at the chromosomal level with those in other model fish species. However, local chromosome shuffling and rearrangements are extensive (Figure 3). Our whole genome comparative analysis with four teleost species also indicated extensive inter-chromosomal rearrangements during evolution, consistent with the hypothesis that inter-chromosomal rearrangements were increased after whole genome duplication in the teleost lineage [67, 72].