Any suggestions for comparing groups of quasispecies clones looking for selection?

Many papers use raw dN/dS data but this seems at best a crude estimate given that the selection either for or against minor lineages may be underestimated.

Sliding window analysis also may be prone to this.

I have been using the codon based z test of selection in MEGA5 but feel that this again does not adequately consider the quasispecies nature of my data.

Any suggestions with regard to software or methodology greatly appreciated