Weizmann Institute of Science, Rehovot, 76100, Israel
csiosch@dapsas1.weizmann.ac.il
As has been demonstrated earlier, the main feature of the nucleosome DNA sequence pattern is periodic distribution of AA (TT) dinucleotides along the sequence. The periodic distribution of CC (GG) dinucleotides along nucleosome sites has been recently demonstrated as well. Since the periods of these distributions are within error bars the same, close to 10.3 bp, one can assume that the periodicity would be the main feature of some other dinucleotides' patterns as well. For evaluation of the contributions of other dinucleotides in the nucleosome pattern an original resonance analysis of our nucleosomal database (204 sequences) was applied. In this procedure the sequences were aligned versus harmonic distributions with the period of 10.3 bp, and the summed output patterns compared to similar outputs generated for random sequences.
That is the sequences were aligned versus the matrix [K*145], where the 145 columns correspond to the nucleosome DNA length, and the K rows - to the input harmonic distribution of K chosen dinucleotides. Each string of the matrix symbolizes a positional distribution of a corresponding dinucleotide and is represented by harmonic function. Varying period and phase shifts between the functions we maximized the sum of the amplitudes of the oscillating components of the output dinucleotide distributions. The AA (TT) dinucleotides were confirmed as main contributors. The amplitudes for CC and GG distributions were found to be also among the strongest as expected. Next candidate pair of the dinucleotides for which nucleosome signal is apparent from the resonance analysis are AC (GT). Thus, AC (GT) dinucleotides were incorporated into the next cycle of optimization (resonance) analysis. The resulting matrix [6*145] constructed as described above is currently used for further cycles of refinement of the nucleosome pattern.