Changes to algorithms since Firth & Brown, 2005, Bioinformatics, 21, 282-92:

• The divergence parameter t is now fitted to make the expected number of mutations under the model equal to the observed number of mutations. Previously, t was fitted to maximize the probability of the observed mutations under the model.
  
  
• A non-coding model has been added to the single-coding and double-coding coding models so it is now easy to analyse query CDSs that overlap other CDSs or non-coding regions or both.
  
  
• Codons that cross gapped regions in the alignment are no longer skipped.
  
  
• Ambiguous nucleotide codes are now allowed in the input sequences.
  
  
• The required format of the input sequences and CDS annotation has been simplified.
  
  
• Circular genomes are fully supported.
  
  
• The method for summing over a multiple sequence alignment has been introduced (previously only pairwise alignments were used).