translate_aln_aa_to_codon.py 文件源码

def main():
    '''
    Back translate aligned amino acid sequences to nucleotide sequences while maintaining the alignment
    '''
    #creating a parser
    parser = argparse.ArgumentParser(
            formatter_class=argparse.RawDescriptionHelpFormatter,
            description='Translate amino acid sequences into nucleotide sequences while maintaining the amino acid alignment and the order of sequences.')
    #adding arguments 
    parser.add_argument('-a', metavar='<aa_aln.fasta>', type=str, 
    help='input amino acid alignment file')
    parser.add_argument('-n', metavar='<nuc_aln.fasta>', type=str, 
    help='input unaligned/aligned nucleotide/codon sequence file')
    parser.add_argument('-o', metavar='<codon_aln.fasta>', type=str, 
    help='output codon alignment file') 
    args = parser.parse_args()

    #set up output file name if none is given
    if args.o is None:
        nuc_aln_file = "codon_aln.fasta"
    else:
        nuc_aln_file = args.o

    aa_aln_file = args.a
    nuc_seq_file = args.n

    aa_aln = AlignIO.read(aa_aln_file, "fasta") 

    nuc_seq = open(nuc_seq_file) #read in codon file
    codon_dict = SeqIO.to_dict(SeqIO.parse(nuc_seq, "fasta")) #read in codon sequence file as a dictionary, key=sequence ID, value=sequence itself

    back_translate(aa_aln,codon_dict,nuc_aln_file,gencode)