First, we authored first alignments of the amino acid sequences to fix prospective frameshifts inside our dataset