Fig. 4

Dinucleotide orientational asymmetries in gene upstream regions. For five upstream regional intervals of length 100 nt (−500..-401, −400..-301,…,-100..-1), logarithmic (base 2) dinucleotide orientation ratios (DORs, see Methods, Eq. 4) of the observed-vs-expected frequency ratios of all possible dinucleotides of its forward relative to the respective reverse-complement version are plotted. Observed-vs-expected frequency ratios measure the departure of actually observed dinucleotide frequencies versus their estimated frequencies based on single base frequencies alone; i.e., treating them as independent events (see Methods for details). Thus, deviations from zero indicate evidence of conditional probability differences between the forward and reverse-complement dinucleotide version and are indicative of orientational preferences. For palindromic dinucleotides (AT,CG,GC,TA), this log-ratio computes as zero. Ratios are plotted for dinucleotide combinations with pairs constituting respective inverse ratios (e.g., TC/GA and GA/TC) necessarily resulting in symmetric graphs