Investigating selection on viruses: a statistical alignment approach.

Area of Science:

• Genomics
• Evolutionary Biology
• Bioinformatics

Background:

• Viral genomes present challenges for selection studies due to overlapping reading frames and high mutation rates.
• Selection in one reading frame can bias neutral mutation rate estimates in another.
• Accurate genome alignment is difficult due to high mutation rates.

Purpose of the Study:

• To develop a model that explicitly accounts for selection in overlapping reading frames.
• To integrate this model into a statistical alignment framework for accurate selection estimation.
• To address the uncertainty associated with genome alignments.

Main Methods:

• Developed a novel model to explicitly model selection in overlapping reading frames.
• Integrated the selection model into a statistical alignment framework.
• Performed simulation studies to evaluate the model's performance against fixed alignments.

Main Results:

• The proposed method yields unbiased selection parameters for viral genomic regions.
• Integrating alignment uncertainty significantly increases estimation accuracy compared to fixed alignments.
• The model performs favorably against established alignment methods like ClustalW and GenAl.
• Applied the method to HIV2 and Hepatitis B sequences, revealing insights into their selection pressures.

Conclusions:

• Marginalizing over all possible alignments is crucial for parametric inference from uncertain sequence data.
• Overlapping coding regions in HIV2 show less stringent selection than single coding regions.
• Evidence suggests differential selection pressures in overlapping reading frames, with one under positive and the other under negative selection.