Chaos Game Representation: An Alignment-Free Technique for Exploring Evolutionary Relationships of Protein Sequences

Abstract

Chaos Game Representation (CGR) is an iterative mapping technique, which shows patterns in amino acids or nucleotide sequences. Here we present a method for using CGR to explore evolutionary relationships of protein sequences based on amino acid properties and illustrate the approach with complete sets of protein translations from viral genomes. In an analysis of complete polyprotein sequences from the viral family Flaviviridae, the CGR method was able to cluster members of major viral groups together, but relations within groups were not well resolved in comparison to an alignment-based phylogeny. We applied the method to members of five different families of ssRNA positive-strand viruses, and each family formed a distinct cluster. We also present a method of testing the reliability of clustering in CGR-based trees, which involves the use of multiple random starting points for the CGR process.