Exploring nonlinearity to identify genes and intergenic regions in genomes

In a previous paper (Physica A 312 (2002) 458) we presented an analysis, which established that DNA sequences exhibit significant nonlinear structure. More specifically, we found that both coding and non-coding sequences exhibit nonlinearity, but coding sequences are more nonlinear than non-coding sequences. Here, we extend this result and we propose an approach which explores nonlinearity to predict genes and intergenic regions in sequenced genomes. Our results compare favorably to other commonly used approaches in this problem.