I think PAML package does have a simple method proposed by Nei & Gojobori
which is not maximum likelihood based in codenml program.
"Robert Querfurth" <Robert.querfurth at gmx.de> wrote in message
news:ct3eor$u9n$1 at news1.zdv.uni-mainz.de...
>> Are there any free programms to infer dn and ds, that are not maximum
> likelihood based.