Dear Yer,

Quote:


HyPhy does not directly implement the same branch-site analyses as PAML (although, if you are interested, I can probably put one together in a few hours).

However, HyPhy has a few of (unpublished) analyses to address similar questions.

In 'Standard Analyses' under 'Positive Selection' there are three relevant analyses:

(1). TestBranchDNDS.bf allows you to test whether the mean dN/dS along a branch (or branches) you select is different from the rest of the tree. You can select from a large variety of codon models and site-to-site rate variation models to test this. This test simply checks if the mean dN/dS is different from the rest of the tree, and does not infer which sites are under selection.

(2). You can use TestSubtreeSelection.bf to apply a fixed effects model (see Multimedia File Viewing and Clickable Links are available for Registered Members only!!  You need to Login for a description of a fixed effects type model) to test whether dN/dS of a given clade (or internal branches) is substantially different from the rest of the tree. This test is done site-by-site

(3). You can also test for positive selection (e.g. compute a p-value that dN>dS along a given set of branches at every site, allowing the rest of the tree to do whatever it wants) along an arbitrary subset of tree branches at a given site using fixed effects likelihood with QuickSelectionDetection.bf. QuickSelectionDetection.bf has many options, the choices you need to make are:

...
...Estimate (for dN/dS bias parameter)
...Two rate FEL (for Ancestor Counting Options)
...
...Custom Subset (for Branch Options)

You may also want to check out Multimedia File Viewing and Clickable Links are available for Registered Members only!!  You need to Login for a genetic algorithms approach to allocating branches to various dN/dS classes. This is also implemented at Multimedia File Viewing and Clickable Links are available for Registered Members only!!  You need to Login

Cheers,
Sergei