Skip Search and Navigation

University of Louisville Bioinformatics

Bioinformatics Research at the University of Louisville


Discovering Expression Patterns and Evolutionary Processes using Bayesian Methods


Dazhuo Li(1,2) Patrick Shafto(3) Eric C. Rouchka(1,2)

  1. University of Louisville Department of Computer Engineering and Computer Science.
  2. University of Louisville Department of Bioinformatics Research Group.
  3. University of Louisville Department of Psychology.

The topologies of phylogenetic trees often disagree when different genes (or proteins) are used to construct the trees. Incongruence between phylogenies indicates two possibilities: (1) the genes all share the same history and the incongruence is due to incorrect model estimation, (2) different genes have experienced distinct evolutionary histories (i.e. via horizontal gene transfer). In either case, phylogenetic analysis based on concatenated genes or consensus methods is problematic due to the lack of information about congruence among genes a priori. Several methods have been developed to assess congruence among markers in phylogenetic analysis, including parsimony-based and maximum likelihood based methods . However, these testing methods are sensitive to rates of variation across sites and to estimated peak likelihood. A Bayesian method explicitly testing congruence has been presented recently. This method, although promising in assessing pairwise gene incongruencies, does not explicitly address multiple gene congruence and partitioning. Motivated by the shortcomings of existing methods, we developed a model that uses Bayesian hierarchical clustering to detect congruence in multigene or multiple protein data sets. The pairwise gene congruencies are based on the posterior probability of two genes having experienced similar evolution history. Building on the pairwise gene congruencies, a systematic approach is proposed to assess and merge multiple genes with relatively high congruence.



  • Li D, Rouchka EC, Shafto P. (2010) "Phylogenomic analysis using Bayesian congruence measuring." in Proceedings of the Second International Conference on Bioinformatics and Computational Biology (BICoB-2010). pp. 30-37. March 24-26, 2010, Honolulu, HI.

Top of Page