• HOME
• PUBLICATIONS
• CURRICULUM VITAE

|Abstract|

Manuel Dehnert, Heike Hameister, Werner E. Helm, Marc-Thorsten Hütt
Transposable elements explain differences and similarities in the correlation structure of Eukaryotic genomes

Attempts to characterize a species on the basis of statistical properties of its DNA sequence have been formulated for several decades. The most prominent of such genome signatures rely on neighborhood correlations (e.g., dinucleotide frequencies) and, consequently, attribute species identification to mechanisms operating on the dinucleotide level (e.g., neighbor-dependent mutations).

In two recent studies we investigated longer-range correlations (up to distances of a few tens of nucleotides) in DNA sequences with methods from information theory. We find that these correlation profiles, when analyzed for a variety of eukaryotic species, display a high degree of intra-species similarity and systematic inter-species differences. Remarkably, these inter-species differences increase with evolutionary distance, i.e. a cluster tree based upon distances of the correlation profile sorts all chromosomes involved into species clusters and approximates the corresponding phylogenetic tree of these species. When comparing closely related species we furthermore find that species distinction on the basis of correlation profiles increases with an increased range of correlations taken into account (up to a few hundreds of nucleotides).

Here we show that the patterning of eukaryotic genomes by repetitive elements explains most (but not all) of these statistical features. We use Repbase data to mask different classes of repetitive DNA in chromosome sequences of different Eukaryotes and observe, how the resulting correlation patterns change. We relate these changes with known differences in properties of mobile genomic elements between the species. Understanding these statistical features as signatures of repetitive elements provides a link to processes of genome evolution, particularly retrotransposition, which pattern the genome on an evolutionary time scale.

By linking the elementary process-oriented properties with genome-wide statistical patterns, this view of the correlation profile as a process signature points towards a system biology treatment of genome evolution.

Heister.org