ATomiCG: A Novel Segmentation Algorithm for Studying DNA Composition

In the last decades, the concept of genome is deeply changed. The application of molecular techniques to the study of whole genomes has had a dramatic impact on the various disciplines of biology. This modern insight of the genome requires the development of new parameters to describe its features. A standard parameter commonly used to describe the DNA composition is represented by the CG content. Based on the CG content, the “theory of isochore” was formulated. However, some objections raised to this theory and a new less drastic definition of “homogeneous domains” was suggested. Several approaches have been proposed for studying the compositional variability of genomes. A segmentation algorithm recently proposed, IsoPlotter, seems to provide a more accurate DNA sequence fragmentation. However, a clear definition of homogeneous domains is still missing.

Using the chemical-physical concept of homogeneous system, we can define a DNA sequence as homogeneous if each its own elementary tiles have the same features (ie the CG content). Based on this definition and using the χ2 test, a new segmentation algorithm (ATomiCG), was used to analyse different genomes. A comparison between ATtomiCG and Isoplotter on simulated data was performed to test the accuracy and reliability of the proposed algorithm. Results highlighted that ATomiCG is a good method able to: split DNA sequence in classes of different CG content; confirm the lack of long fairly homogeneous domains and find a correlation between CG content and gene density.