Phylogenetic inferences from molecular sequences: Review and critique

Conflicting results often accompany phylogenetic analyses of RNA, DNA, or p rotein sequences across diverse species. Causes contributing to these confl icts relate to ambiguities in identifying homologous characters of alignmen ts, sensitivity of tree-making methods to unequal evolutionary rates, biase s in species sampling, unrecognized paralogy, functional differentiation, l oss of phylogenetic informational content due to long branches or fast evol ution, and difficulties with the assumptions and approximations used to inf er phylogenetic relationships. Attempts to surmount these conflicts by aver aging over many proteins are problematic due to inherent biases of selected families, lack of signal in others, and events of lateral transfer, fusion , and/or chimerism. The process of assessing reliability of the results usi ng the bootstrap method is strewn with obstacles because of lack of indepen dence and inhomogeneity in the molecular data. Problems inherent to the thr ee major procedures for developing phylogenetic trees-parsimony, likelihood , distance-are reviewed. Special attention is given to the problem of infer ring evolutionary distances from patterns of similarity among sequences. Th e difficulties encountered by methods of phylogenetic reconstructions based on the analysis of divergent sequence families make new methods based on t he analysis of complete genomes reasonable alternatives. Several of these a re considered, including the signature sequences of Gupta and associates, t he study of genome profiles, and the genomic signature set forth by Karlin and colleagues. (C) 2001 Academic Press..