Hello,
For three alignments I've tried, the estimate of dN/dS fromquickselectiondetection.bf is very different from dN/dS estimate using dSdN.bf (from phylohandbook section) - using the same tree + MG94xRev. E.g. for one gene values are 0.66 vs. 0.44. For all 3 cases, the estimate from quickselection was greater than from dSdN.
I also tried both methods on MammalsBirds.nex from phylohandbook data
- using quickselectiondetection.bf (with MG94xRev): dN/dS=0.0931387
- using dSdN.bf: dN=0.124807, dS=3.27109 (dN/dS=0.038)
It is not clear to me how these methods differ - from hyphybook2007 and my previous understanding it seems like:
- dN/dS in quickselection is the MLE of beta/alpha given the codon alignment/tree
- dN and dS in dSdN are calculated using expected syn and nonsyn substitutions, which are determined based on the rate matrix given the MLE of all parameters (which would include beta/alpha?)
What have I got wrong?
Thanks very much
Leslie