Capturing whole-genome characteristics in short sequences using a naïve Bayesian classifier

Area of Science:

• Genomics
• Bioinformatics
• Evolutionary Biology

Background:

• Bacterial genomes exhibit distinct nucleotide compositions and oligonucleotide frequency profiles, serving as unique genomic signatures.
• Evolutionary divergence leads to measurable differences in genomic sequences, including GC content and short oligonucleotide patterns.

Purpose of the Study:

• To develop a computational method for predicting the genome of origin for a given DNA sequence based on oligonucleotide frequencies.
• To assess the accuracy of this method for classifying short genomic sequences and identify horizontal gene transfer events.

Main Methods:

• A naïve Bayesian classifier was developed and trained on 28 eubacterial and archaeal genomes.
• The classifier analyzed oligonucleotide frequency profiles to determine the genomic origin of sequences.
• The method was applied to identify horizontal gene transfer events in whole-genome sequences.

Main Results:

• Sequences as short as 400 bases were classified with 85% accuracy.
• The classifier successfully identified the transfer of specific genes (sodC, bioC) between Haemophilus influenzae and Neisseria meningitis, pinpointing donor and recipient species.
• The study validated the utility of oligonucleotide frequency analysis for genomic classification.

Conclusions:

• Oligonucleotide frequency analysis provides a robust method for bacterial genome classification.
• This approach is effective in identifying horizontal gene transfer events, offering insights into microbial evolution.
• The developed classification methodology holds potential as a valuable tool for biodiversity studies and genomic analysis.