gaps as novel stop codons

Hi,

when running treesub on incomplete sequences (i.e. alignment with gaps) stop codons emerge (e.g. I96*) in the output. This happens for entire stretches of sequences.
Is there a way to circumvent this?

Regards,

Thierry