This study presents an improved method for bacterial DNA fingerprinting that addresses two key challenges: selecting appropriate restriction enzymes and efficiently comparing fingerprints across multiple isolates. The researchers use Markov chain analysis of nucleotide frequencies from published DNA sequences to predict restriction enzyme cutting frequencies, eliminating the need for empirical testing. They validate this approach across six bacterial species with varying G+C content (33-75 mol%). For fingerprint comparison, the authors divide electrophoretic gels into sections using 1 kb DNA ladders as reference standards, score band numbers in selected sections to generate numerical profiles, and use these profiles to group similar isolates. The method is demonstrated on diverse organisms including Haemophilus influenzae, Neisseria meningitidis, Staphylococcus aureus, and Aeromonas hydrophila. Results show excellent correlation between predicted and observed restriction frequencies. The sectioning approach allows rapid classification of multiple isolates from single gels, with manual verification of highly similar patterns. This system significantly reduces analysis time and cost compared to traditional fingerprinting methods, making it practical for epidemiological and population genetic studies.

Key findings

• Markov chain analysis of di- and trinucleotide frequencies from DNA sequences accurately predicts restriction enzyme cutting frequencies, correlating well with experimental results across diverse bacterial species
• Restriction enzyme selection is strongly influenced by genome G+C content, with AT-rich recognition sequences cutting more frequently in AT-rich genomes and GC-rich sequences cutting more frequently in GC-rich genomes
• A fingerprint sectioning method using 1 kb DNA ladder standards allows rapid numerical profile comparison of multiple isolates from single electrophoretic gels, reducing time and cost of bacterial strain characterization
• The technique successfully groups related isolates and discriminates strain variants, with application demonstrated in non-typable Haemophilus influenzae and other clinically relevant bacteria