Hi,
If you have two data partitions from one alignment can you use the estimated base frequence from one partition to estimate parameters for the other partition?  The logic behind this is that we only want to analyze four fold degenerate sites from the total data so we've written a program to identify these sites - but rather than individually choose each site it would be easier to put these sites at the end of the sequence alignment  and thus partition the alignment and the four fold sites separately.  Thanks!

Dear Jennifer,

Let me make sure I understand your question correctly.

1). Your alignment is partitioned (a priori) into two classes of sites (class A and B).
2). You want to use character frequencies from sites in class A, to do inference on partition B.

That is definitely possible (one has to write a custom batch file, but it's not too tedious and I can give you a starting point), but in order to provide you with a helpful answer I need more information. Specifically:

a). Do you wish to use nucleotide or codon models?
b). Do you want to do joint inference on partitions (optimize them together, but using shared base frequencies), or do you want to first analyze partition A, then pipe some of the output as input to a separate analysis of partition B?
c). When you say 'estimate parameters', what parameters do you mean: e.g. branch lengths and substitution biases, only substitution biases, etc.

Cheers,
Sergei