RESUMO
BACKGROUND: As the number of fully sequenced genome increases, the need is greater for bioinformatics to predict or annotate genes of a newly sequenced genome. Ever since Eisenberg and his colleagues introduced phylogenetic profiling for assigning or predicting protein functions using comparative genomic analysis, the approach has been used in predicting function of some prokaryotic genomes quite successfully. Very little work has been reported in functional prediction of eukaryotes such as mouse and Homo sapiens species from phylogenetic profiles. RESULTS: We have proposed a general methodology for validating the hypothesis underlying phylogenetic profiling techniques, and have demonstrated it using eukaryotic target genomes such as Homo sapiens and mouse. The gene ontology is used as the gold standard for validating functional similarity among the genes in each cluster. We compute the functional cohesiveness of each cluster and the results appeared to be not encouraging towards finding functionally cohesive phylogenetic profiles. This result complements one recent work on the poor performance on functional linkage in some eukaryotic genome using phylogenetic profiling techniques. If we introduce a broad interpretation for functionally related genes as functional sub-clustering within a phylogenetic profile, then we have a very strong support for the hypothesis as we have shown in the paper.
