Numerical characterization of DNA sequences: connectivity type indices derived from DNA line graphs

Abstract

The four-letter code sequence of a single strand of a DNA sequence was converted into a line graph, and the vertices of the line graph were assigned weights according to the dissociation constant (pKa) of the corresponding nitrogenous base represented by each of the vertices. Connectivity type indices were computed for the weighted line graphs and the numerical descriptors thus calculated were used for alignment-free sequence comparison. The numerical descriptors proposed in this study were calculated very fast even for whole genomes, and thus, the methodology enabled alignment-free comparison of long DNA sequences without much computational load. Sequence comparison using numerical descriptors derived from the weighted line graphs is illustrated using 23 mitochondrial genomic sequences. The cladogram obtained from the hierarchical clustering carried out using the numerical descriptors grouped evolutionarily similar sequences together.