Hi again, HyPhyers,

Still posting about the same problem - implementation of substitution models for a binary character (i.e. with two possible states, 1 and 0).

I'd like to implement three models
[list][*]both rates are equal i.e. q[sub]10[/sub] = q[sub]01[/sub]
[*]rates are unequal, but model remains reversible i.e. q[sub]10[/sub] =/= q[sub]01[/sub] but pi[sub]1[/sub] * q[sub]01[/sub] = pi[sub]0[/sub] * q[sub]10[/sub]
[*] rates are unequal and model is nonreversible i.e. q[sub]10[/sub] =/= q[sub]01[/sub] and pi[sub]1[/sub] * q[sub]01[/sub] =/= pi[sub]0[/sub] * q[sub]10[/sub]
[/list]
At the moment, I use
[code]binaryNonRevQ = {{*, forwardRate*t}
                {backwardRate*t,*}};[/code]
for what I think/thought was a nonreversible Q, while adding
[code]forwardRate:=backwardRate;[/code] to get the reversible model

However, I'm now not sure, when I implement the initial binaryNonRevQ, whether I'm actually just implementing an asymmetric but still reversible (i.e. where pi[sub]1[/sub] * q[sub]01[/sub] = pi[sub]0[/sub] * q[sub]10[/sub]) model - and if so, how should I get a non-reversible model?  

Or is, indeed, the model I describe non-reversible, and I'd need to somehow specify something like q[sub]01[/sub] := pi[sub]0[/sub] * q[sub]10[/sub] to get the asymetric-rate reversible model? (If so, a hint on how to do that would be great).

Thanks

Aidan