A standard deviation based quantification differentiates coding from non-coding DNA sequences and gives insight to their evolutionary history.

A method quantifying the randomness of nucleotide sequences is developed, based on the introduction of a standard deviation type of quantity involving locally computed means and a length scale around which is assessed the clustering of nucleotides. It is pointed out that the value taken by this modified standard deviation may distinguish between coding rich and non-coding rich sequences. Moreover, the approach described herein allows the determination of some minimal characteristics of an evolutionary scenario which can account for the origin of the clustering in the nucleotide distribution of the different parts of the genome.