Reconstructing the History of Yeast Genomes 英文参考文献.docVIP

Perspective ReconstructingtheHistoryofYeastGenomes DavidSankoff* UniversityofOttawa,Ottawa,Ontario,Canada Some 12 years ago, Wolfe and col- leagues demonstrated that Saccharomyces cerevisiae is the descendant of an ancient whole-genome duplication event [1,2], much to the consternation of many of those who had recently completed the sequencing of this yeast [3], the first eukaryotic nuclear genome to be se- quenced. Despite persistent rejectionist argument[4],thisbreakthroughdiscovery has been amply confirmed [5,6] and has beenthestartingpointforscoresofpapers on yeast evolution and phylogeny, culmi- nating in the Yeast Gene Order Browser [7]andthepaperbyGhis issueofPLoSGenetics[8]. Conceptually,thephylogeneticstudyof gene content, including gene gains and losses, does not depend on gene order considerations.Indeed,apreliminarystep in the method of Gordon et al. is the inference of the gene content at the ancestral nodes of the assumed phyloge- netictreeof11yeastspecies.Sincespatial proximityoffunctionallyinteractinggenes on chromosomes is relatively less impor- tant than in prokaryotes, the evolution of quence represents but a small proportion of the genome, have been constructed based on banding patterns [9], genomic sequence[10],andeverythinginbetween, delimitationofthesubtelomericregionsto be excluded from the analysis required highly informed scientific judgment to make the trade-off between increased for high-resolution analyses, complete coverage and increased uncertainty. sequences,includingtherelativelyrapidly evolving intergenic regions, should be used. For gene-dense eukaryotic genomes Third, computer programs suffer from both simplistic objective functions and overly constrained models of gene order such as those of Drosophila [11] or Saccha- change, both of which can lead to romyces [8], however, gene order data represent the best compromise between maximum coverage of the genome and maximum confidence in the orthology identifications. m