A classification approach for genotyping viral sequences based on multidimensional scaling and linear discriminant...

Area of Science:

• Bioinformatics
• Virology
• Computational Biology

Background:

• Accurate viral genotype classification is crucial for understanding viral evolution.
• Existing methods may not fully leverage comprehensive sequence data.
• Divergent viruses like HIV-1 and HCV require robust classification tools.

Purpose of the Study:

• To introduce MuLDAS, a novel computational approach for classifying viral genotypes.
• To develop a method utilizing statistical genotype models learned from known sequences.
• To provide a tool for accurate genotype assignment in rapidly evolving viruses.

Main Methods:

• MuLDAS aligns query sequences to reference alignments and computes a distance matrix.
• Multidimensional scaling maps sequences to a principal coordinate space.
• Linear discriminant models are trained on reference sequence coordinates for genotype partitioning.

Main Results:

• MuLDAS achieved high concordance rates: 99.3% for HIV-1 and 96.6% for HCV.
• The method provides confidence estimates and detects outlier sequences.
• Performance was validated against benchmark datasets from Los Alamos National Laboratory.

Conclusions:

• MuLDAS offers highly accurate genotype assignment for rapidly evolving viruses.
• The inclusion of evaluation measures enhances its utility in viral sequence analysis.
• A web server is available for public access to the MuLDAS tool.