(The) correlation between GC contents and the frequencies of dinucleotides in DNA sequences

Abstract

Unique DNA sequences of various organisms were analyzed to understand the relationship between dinucleotide frequencies and GC contents. In general, dinucleotides in DNA sequences are over- or under-represented in comparison with the random or ideal sequences. DOEs (Observed per Expected Frequencies of Dinucleotides) were calculated and seemed to correlate with GC contents positively or negatively and generally have different values according to GC contents. DOE-GC distributions were speciesspecific and showed general trend. This trend divided organisms studied into roughly two groups, vertebrates and non-vertebrates. The dinucleotide-driven mutation pressure of D. melanogaster (the fruitfly) seems to be similar to that of E. coli. Also, the relation-ship between DFs (Frequencies of Dinucleotides) and GC contents was studied. Random-model for DF-GC distribution was proposed and with this model and DOE, human-model was predicted and evaluated to be correct. In general, there is the relationship, DF(x) = DOE(x)$\cdot$DF$_e$(x), in the natural DNA sequences, where x is the GC fraction, and DF$_e$(x) is mathematically expected DF.