I have added a standard analysis (builds on Feb 8th or later) which compares the fits of MG94 and GY94 models crossed with an arbitrary nucleotide substituion bias matrix on a codon alignment.

Some of the features:

1). Use exact or approximate branch lengths
2). Tabulate how many sites support MG94 and how many GY94
3). Perform a KH resampling test to assess sampling influence
4). Obtain plots (Mac and Windows versions) of nucleotide and codon frequencies (F1x4, F3x4 and empirical estimators) and site-by-site log likelihoods.
5). Derive profile likelihood confidence intervals on the average dN/dS value.

Cheers,
Sergei