STATISTICAL-ANALYSIS OF L-TUPLE FREQUENCIES IN EUBACTERIA AND ORGANELLES

This work is an attempt to study the structural features and evolution ary patterns of nucleotide sequences by analyzing their 1- through 4-p let frequencies and statistical relations between them. We present mat hematical apparatus for this analysis. In particular, we introduce cri teria to estimate the degree of homogeneity of L-plet composition in a given set of sequences and the dependence of the L-plet frequencies o n the composition of lower orders. We apply these criteria to the stud y of eubacteria, mitochondria and chloroplasts. We demonstrate that L- plet frequencies are quite useful for revealing evolutionary relations hip between DNA sequences and that the non-random distribution is more typical for doublets than to triplets. Non-randomness of triplet comp osition is more characteristic to coding than to non-coding regions, w hile no significant differences in dinucleotide composition can be obs erved. The obtained results can be used for revealing possible mechani sms of the codon usage phenomena.