Search

We show that the nearest neighbour distribution of distances between basis pairs of some intron-less and intron-containing coding regions are the same when a procedure, called {\em unfolding}, is applied. Such a procedure consists in separating the secular variations from the oscillatory terms. The form of the distribution obtained is quite similar to that of a random, i.e. Poissonian, sequence. This is done for the HUMBMYH7CD, DROMYONMA, HUMBMYH7 and DROMHC sequences. The first two correspond to highly coding regions while the last two correspond to non-coding regions. We also show that the distributions before the unfolding procedure depend on the secular part but, after the unfolding procedure we obtain an striking result: all distributions are similar to each other. The result becomes independent of the content of introns or the species we have chosen. This is in contradiction with the results obtained with the detrended fluctuation analysis in which the correlations yield different results for intron-less and intron-containing regions., Comment: 8 pages. elesart style.Alberto Robledo's 60th Anniversary Symposium - Nonlinearity, nonequilibrium and complexity: Questions and perspectives in statistical physics. Accepted in Physica A

a b s t r a c t Statistical analysis of whole genomic sequences usually assumes a homogeneous nucleotide density throughout the genome, an assumption that has been proved incorrect for several organisms since the nucleotide density is only locally homogeneous. To avoid giving a single numerical value to this variable property, we propose the use of spectral statistics, which characterizes the density of nucleotides as a function of its position in the genome. We show that the cumulative density of bases in bacterial genomes can be separated into an average (or secular) plus a fluctuating part. Bacterial genomes can be divided into two groups according to the qualitative description of their secular part: linear and piecewise linear. These two groups of genomes show different properties when their nucleotide spacing distribution is studied. In order to analyze genomes having a variable nucleotide density, statistically, the use of unfolding is necessary, i.e., to get a separation between the secular part and the fluctuations. The unfolding allows an adequate comparison with the statistical properties of other genomes. With this methodology, four genomes were analyzed Burkholderia, Bacillus, Clostridium and Corynebacterium. Interestingly, the nearest neighbor spacing distributions or detrended distance distributions are very similar for species within the same genus but they are very different for species from different genera. This difference can be attributed to the difference in the codon usage.