Motivation: A number of methods for detecting positive selection in protein coding DNA sequences are based on whether each site/region has a non-synonymous to synonymous substitution rates ratio ω greater than one. However, a site/region may show a relatively large number of non-synonymous mutations that conserve a particular property. Recent methods have proposed to consider as evidence for molecular adaptations how conserving, or radically different, non-synonymous mutations are with respect to some key amino acid properties. While such methods have been useful in providing a qualitative assessment of molecular adaptation, they rely on independent statistical analyses for each amino acid property and typically do not properly adjust for multiple comparisons when selection needs to be assessed at several sites. Results: We consider a Bayesian hierarchical model that allows us to jointly determine if a set of amino acid properties are being conserved or radically changed while simultaneously adjusting for multiple comparisons at the codon level. We illustrate how this model can be used to characterize molecular adaptation in two datasets: an alignment from six class I alleles of the human major histocompatibility complex and a sperm lysin alignment from 25 abalone species. We compare the results obtained with the proposed hierarchical models to those obtained with alternative methods. Our analyses indicate that a more complete quantitative and qualitative characterization of molecular adaptation is achieved by taking into account changes in amino acid properties.
ASJC Scopus subject areas
- Statistics and Probability
- Molecular Biology
- Computer Science Applications
- Computational Theory and Mathematics
- Computational Mathematics