An efficient numerical representation of genome sequence: natural vector with covariance component

Area of Science:

• Genomics and Bioinformatics
• Microbial Ecology
• Evolutionary Biology

Background:

• Comparing microbial sequences (archaea, bacteria, viruses, fungi) is crucial for evolutionary and population studies.
• Existing methods often face limitations in sequence length and generality.
• A need exists for a universal method for microbial sequence characterization and phylogenetic analysis.

Purpose of the Study:

• To propose a general, alignment-free method for microbial sequence characterization.
• To analyze the classification and phylogenetic relationships within microbial datasets.
• To investigate the evolutionary placement of giant viruses relative to other microorganisms.

Main Methods:

• Developed a novel alignment-free method representing biological sequences using an 18-dimensional natural vector.
• Incorporated covariance between nucleotides to enhance sequence representation.
• Applied the method to analyze 24,250 genomic and 95,542 DNA barcode sequences.

Main Results:

• Validated the necessity of a six-dimensional covariance vector for accurate sequence characterization.
• Demonstrated giant viruses exhibit nucleotide distribution similarities closer to bacteria.
• Phylogenetic analysis revealed Mimiviridae clustered with Pandoraviridae, with Mimiviridae closer to the root than Marsellieviridae.

Conclusions:

• The 18-dimensional natural vector provides an effective numerical representation for microbial sequences.
• The alignment-free method is computationally efficient and suitable for large-scale genomic analysis.
• Findings contribute to a refined understanding of microbial evolution, particularly the origins of giant viruses.