Robust phylogenetic profile clustering for proteins.

BACKGROUND

Genes are continually formed and lost as a genome evolves. However, new genes may tend to appear during specific evolutionary epochs rather than others, or disappear together in a more recent organismal clade. Methods to identify gene origination might simply use the last common ancestor to contain an ortholog as the putative gene origination point, or use a heuristic threshold that allows for a certain amount of missing orthologs in the cohort of species examined. Here, to avoid such issues, an alternative approach based on the clustering of phylogenetic profiles is applied, and the results are examined for any evidence of epochal trends in gene origination, and associated trends in specific sequence traits or functional associations.

METHODS

A phylogenetic profile is simply an array indicating the presence or absence of a gene in a list of species. These profiles were compared and clustered to discern patterns in gene occurrences across >800 fungal species, centering the analysis on the budding yeast .

RESULTS

Clear epochs of gene origination were observed linked to the last common ancestors of and , and also to and earlier ancestors. These trends are independent of the proteome and genome-assembly quality of the underlying data. Clusters of phylogenetic profiles demonstrated some significant functional associations, such as to cellular spore formation and chromosome segregation in genes originating in . The phylogenetic profile clustering analysis enabled detection of parameter-independent trends in intrinsic disorder, prion-like composition and gene uniqueness as a function of epochal gene age. For example: new proteins with prion-like domains have arisen at a similar rate for most of fungal evolution centred on ; the most proteins with mild intrinsic disorder have appeared during the early epoch rather than more recently, and very recently formed genes are the least likely to be single-copy (., 'unique' yeast proteins).

CONCLUSIONS

For individual proteins, the profile cluster data generated here are useful for investigating experimental hypotheses, since they provide evidence for functional linkages that have yet to be discerned.