You can try HyPhy (http://hyphy.org/w/index.php/Main_Page) or the DataMonkey server (http://www.datamonkey.org/).

For selection in non-coding sequences, try to have a read at this http://genome.cshlp.org/content/20/5/685.short

Also you can try GERP++, PhastCons, Siphy-w and Siphy-pi.

OmegaMap uses a population genetics Bayesian framework and is considerably simpler to implement that PAML

I use the Selecton program which can be run on the server or downloaded to run on your own machine. There are a number of codon models available there but I prefer to use the Mechanistic + Empirical Codon model (MEC).

Paper:

http://www.ncbi.nlm.nih.gov/pubmed/15647294

Site:

selecton.tau.ac.il/

I strongly recommend MEGA6. It simplifies the entire phylogenetic reconstruction process (for nucleotides and codons, proteins, etc) and narrows the models relevant to the reconstruction model you wish to use. It is menu driven, and well documented. You may need to dig into the literature to learn the ins-and-outs of particular evolutionary models, or follow examples from other papers. http://megasoftware.net You can thank me later :)

I used MEGA mostly for phylogenetic trees, but thank you Nick, I will try this too.