Hi Sergei,

The documentation states that one can use HyPhy with irreversible models, and indeed such a model is easy to specify in HyPhy. However, they only make sense if used with rooted trees - the likelihood very much depends on the position of the root.

From GettingStarted.pdf: "however, for most purposes rooted trees are automatically unrooted by HyPhy because likelihood values for unrooted trees are the same as those for rooted trees."

Is it possible to force HyPhy to use rooted trees? Or implement irreversible models?

If a general solution is not available, there may still be one for my particular case: I want to use one specific taxon as the root of the tree. With a rooted tree this could be achieved by constraining the appropriate branch length to zero. With an unrooted tree it may still be possible: if HyPhy starts its likelihood calculation at a specific node, then that node is effectively the root of the tree and all one needs is to know how HyPhy chooses the starting node.

cheers,
Konrad

Thanks Sergei,

I found NRM.mdl but not NRV.bf

Setting the flag changes the tree data structure, but I am uncertain how it works. For instance, using the following tree from my data file:

(tax1:0.1,tax2:0.1);

Directly outputting the tree data structure using fprintf gives:

(tax1:0,(tax2:0)Node2:0);

(Why is it ignoring the specified branch lengths?)

On changing the second branch length using givenTree.tax2.t := 10;  this changes to:

(tax1:0 (tax2:0.120039)Node2:0);

Could you explain how the .t parameter relates to the branch length? Also, is the root of the tree at Node2, or at the point where the branch from Node2 joins with the branch from tax1?