Detecting and characterizing gene conversions between multigene family members

We used a variety of methods to detect known gene conversions in the actin gene families of five angiosperm species, the beta-globin gene families of two primate species, and the Zfx/Zfy gene families of seven mammalian species. Our goal was to devise a working strategy which would allow the analysis of the members of a multigene family in order to determine whether there had been gene conversions between its members, identify the genes involved in the gene conversions, establish the lengths of the converted regions, and determine the polarities of the gene conversions. We show that three phylogenetic methods and the homoplasy test of Maynard Smith and Smith perform relatively poorly on our data sets because the sequences we analyzed had large levels of multiple substitutions. The method of Sawyer, the compatibility method of Jakobsen and Easteal, the partition matrix method of Jakobsen, Wilson, and Easteal, and the co-double method of Balding, Nichols, and Hunt can be used to identify the genes which have been involved in gene conversions. The co-double method is more powerful than other methods but requires orthologous sequences from related species. Compatibility, phylogenetic, and nucleotide substitution distribution statistics methods can be used to identify the location of the converted region(s). Site-by-site compatibility analyses can also be used to identify the direction of the conversion event(s). Combinations of these methods can therefore be used to establish the presence, locations, and polarities of gene conversions between multigene family members.