Dear Jess,

There is a special analysis for doing relative rate tests in HyPhy (under the Relative Rate rubrik in standard analysis). Please use that analysis for relative rate testing.

A relative rate test is effectively a local molecular clock on the (sp1,sp2) clade of the 3-taxon (out,(sp1,sp2)). When you test for a global molecular clock in this situation, you will not be able to distinguish between rate variation in the outgroup and the (sp1, sp2) clade.

HTH,
Sergei

Dear Jess,

Ah - your question makes sense now. HyPhy, unlike most other programs is capable of estimating multiple parameter per branch. For instance, if you have the HKY85 nucleotide model, each branch can have two parameters - the transition rate and the transversion rate (c.f. Multimedia File Viewing and Clickable Links are available for Registered Members only!!  You need to Login). The local model will estimate these two rates separately for each branch, and then you can test for relative rates only on transition rates (this option is offered by the 'which parameters to constrain dialog'). This makes more biological sense when you use codon models - with the local option you can test relative rates on synonymous rates, non-synonymous rates or both.

Global models on the other hand avail you of the standard parameterization - in the case of HKY85 that would mean a single transition/transversion ratio for the entire tree and a single branch length parameter per branch. In this scenario the only parameter you can test is the total branch length.

HTH,
Sergei