Hi,
I have performed a SLAC analysis on a dataset of 80 viral sequences to detect positive selection and I am trying to interpret the results. I got a mean dn/ds of 0.4, which indicates purifying selection. However I got several codons with signals of positive selection.  
How can I explain the different results? How is the mean dn/ds calculated?  Is it an average of all dn/ds values of each codon in the alignment?

Thanks a lot for any help in this matter.

Dear audrinab,

Your question is difficult to answer concisely and informatively. There is no inconsistency in the report that MEAN dn/ds (the entire gene) is < 1, while for some sites it is >1. There is no direct relation between the mean dN/dS reported by SLAC (based on maximum likelihood) and site-by-site estimates of dN/dS (inferred using counting heuristics. Loosely, the mean of SLAC dN/dS (by site) should be similar to the reported mean dN/dS, but the exact relationship is non-trivial.  Please refer to our selection detection book chapter (Multimedia File Viewing and Clickable Links are available for Registered Members only!!  You need to Login) for the necessary information.

Sergei