Models of Protein-coding Gene Evolution

Abstract: Codon substitution models have traditionally been used to measure selective pressures in protein-coding genes by evaluating the ratio of rates of nonsynonymous (amino acid replacing) to synonymous (silent) substitutions. More recent approaches have explored the idea of explicitly modeling site-specific purifying selection at the amino acid level. Loosely speaking, with such models, substitutions at a given position occur at the neutral or near-neutral rate when they do not alter the amino acid encoded, or when they correspond to replacements within a sub-set of suitable amino acids—substitutions to ill-suited amino acids have much lower rates. Two perspectives have been taken with regards to the amino acid variables in such models: in the first, they are treated as random variables integrated over a prior law within the likelihood function; in the second, they are treated as bona-fide parameters, optimized to their ML estimates. The first approach is well-behaved, whereas the latter approach has serious statistical problems. Beyond a discussion of these issues, we explore the idea of using the random variable approach as the null against which to test for deviations from the neutral/nearly-neutral evolutionary regime. We present applications of this idea on a few data sets of protein-coding genes, and discuss how such methods fit within a long-term integrative modeling project. Location: MS 431 Math Science Bld.

Location:

MS 431 Math Science Building

Speaker:

Dr. Nicolas Rodrigue,Dept. of Mathematics and Statistics,University of Calgary

More information at http://www.ucalgary.ca/events/calendar/models-protein-coding-gene-evolution