Parses all proteins from antismash and labels them by genome and cluster number

parse_antismash_proteins.py

import glob
from Bio import SeqIO
for fn in glob.glob('./antiSMASH/*/*cluster*.gbk'):
        genome = fn.split("/")[-2]
        cluster_num = fn.split("cluster")[1].split(".")[0]
        i = 0
        for record in SeqIO.parse(fn, 'genbank'):
                for feature in record.features:
                        if feature.type == 'CDS':
                                print(">" + genome + "|" + cluster_num + "|" + str(i) + "|" + str(feature.location.start) + ":" + str(feature.location.end))
                                print(feature.qualifiers['translation'][0])
                                i += 1
                break