Evolutionary Tree Based on Oligonucleotide Frequencies and Conservative Words in 16S and 18S Ribosomal RNA

Sequence distances are defined in terms of the differences in the occurrence frequencies in sequences of oligonucleotides of length n. Such n-distances are used to construct phylogenetic trees from a set of thirty-five 16S or 18S rRNA sequences. The quality of the trees generally improves with increasing n and reaches a plateau at n=7 or 8. The best n-distance trees are compatible to trees based on sequence alignment, suggesting that highly overrepresented 7-mers and 8-mers are closely related to rRNA evolution. Out of the 47=16384 7-mers, 612 are identified as those whose relative frequencies correlate strongly with the 35×35 n-distance matrix. These evolution-related 7-mers are used to identify “conservative words”, oligonucleotides whose frequencies and loci are common to at least 85% of organisms preselected to represent a domain. The structural meaning of some of these conservative words is discussed.