Jeffrey L. Thorne

Visit Our Lab Home Page

Statistical tools for the analysis of DNA and protein sequences

My colleagues and I study evolution. We do this by developing statistical techniques for analyzing DNA and protein sequence data. Our main efforts concern:

(1) Improving probabilistic models of DNA sequence evolution by incorporating phenotype and reconciling these models with population genetics. The relationship between phenotype and survival of the genotype is central to both genetics and evolution. The field of population genetics has a rich body of theory for explaining how within-species genetic variation is shaped by fitness, mutation, recombination, population size, and population structure. However, this theory does not purport to map genotypes to phenotypes nor does it map phenotypes to fitness. A wide variety of computational biology schemes aim to predict phenotype from genotype. We are working to improve models of molecular evolution by incorporating these computational biology prediction systems. We have concentrated on protein tertiary structure and RNA secondary structure, but are very excited by the potential to quantify the impacts on evolution of diverse other aspects of phenotype. Rather than designing our statistical techniques exclusively for understanding within-species genetic variation, we have been attempting to apply population genetic theory to data sets representing sequences from different species. This is a challenging endeavor but a paucity of intraspecific genetic variation means that many of the most important evolutionary questions can only be addressed via interspecific comparisons.

(2) Evolution of the rate of evolution. Evolutionary analysis of DNA and protein sequences is typically performed by either assuming that all evolutionary lineages change at the same rate or by avoiding any attempt to directly consider the fact that the rate of evolution changes over time. Factors that affect the rate of molecular evolution (e.g., mutation, population size, generation time, selection) change over time and therefore the rate of molecular evolution is extremely unlikely to be identical for different evolutionary lineages. However, it is reasonable to expect an autocorrelation of rates over time. Closely related evolutionary lineages tend to evolve at similar rates and distantly related lineages might evolve at more different rates. My collaborators (especially Hirohisa Kishino of the University of Tokyo) and I are developing methods for estimating dates of evolutionary events from molecular sequence data. These methods lack the restrictive and implausible assumption that rates of evolution have been constant over time. We also feel that these methods have great potential for illuminating patterns of evolutionary rate variation over time.

Selected Publications:

Choi, S.C., Redelings, B.D., and Thorne, J.L. (2008). Basing population and genetic inferences and models of molecular evolution upon desired stationary distributions of DNA or protein sequences. Phil. Trans. R. Soc. B. October 7 [Epub ahead of print].

Choi, S.C., Stone, E.A., Kishino, H., and Thorne, J.L. (2008). Estimates of natural selection due to protein tertiary structure inform the ancestry of biallelic loci. Gene. July 29 [Epub ahead of print].

Thorne, J.L. (2007). Protein evolution constraints and model-based techniques to study them. Current Opinion in Structural Biology. 17: 337–341.

Move directly to the Genetics Home Page.
Move directly to the CALS Home page.

Comments should be sent to theDepartment of Genetics
Genetics Web display last changed 5 November 2008.