Variable-order sequence modeling improves bacterial strain discrimination for Ion Torrent DNA reads.

Area of Science:

• Bioinformatics
• Genomics
• Computational Biology

Background:

• Genome sequencing is vital for pathogen detection and outbreak resolution.
• Accurate DNA read mapping and classification are critical for genomic analysis.
• Current mapping methods often overlook base dependencies, limiting accuracy.

Purpose of the Study:

• To introduce VarHMM, a variable-order paired hidden Markov model for sequence alignment.
• To address practical implementation issues of higher-order Markov models in read mapping.

Main Methods:

• Development of a variable-order paired hidden Markov model (VarHMM).
• Application of VarHMM to DNA read mapping and alignment.
• Comparative analysis against existing alignment methods using Ion Torrent sequenced data.

Main Results:

• VarHMM models higher-order distributions, enhancing alignment probability quantification.
• VarHMM demonstrated superior strain discrimination compared to other methods in tests.
• Improved accuracy in mapping DNA reads to bacterial genomes was achieved.

Conclusions:

• Higher-order probability distribution modeling offers significant advantages for read mapping.
• VarHMM provides more detailed and accurate alignment probabilities.
• Further development of such models holds promise for diverse read mapping applications.