Hi

I'm looking at CMV immune evasion genes, some of which have been shown previously to exist as distinct genotypes within circulating viral populations. Of course, when you look at sequences of genes for which you have no phenotypic data (i.e. you have no idea what the gene does), the questions is how to identify or define a genotype? For example, a NJ tree of 34 sequences of the m04 gene from MCMV looks to me to represent 6 genotypes. But a 'lumper' may think there are only 3 genotypes from the same tree.
Looking at the same sequences using IFEL identifies 10 codons under positive selection on ancestral branches.
So, my long winded question is this: can we assume that positive selection in ancestral sequences are what results in the formation of genotypes, and therefore will an NJ tree of only those codons under positive selection give me a more definitive identification of the number of genotypes within a population?

Toughts anyone?