88770 name: Panarthropoda blast name: Panarthropoda rank: no rank parent: 1206794
1206794 name: Ecdysozoa blast name: Ecdysozoa rank: no rank parent: 33317
33317 name: Protostomia blast name: Protostomia rank: no rank parent: 33213
33213 name: Bilateria blast name: Bilateria rank: no rank parent: 6072
6072 name: Eumetazoa blast name: Eumetazoa rank: no rank parent: 33208
33208 name: Metazoa blast name: animals rank: kingdom parent: 33154
33154 name: Opisthokonta blast name: Opisthokonta rank: no rank parent: 2759
2759 name: Eukaryota blast name: eukaryotes rank: superkingdom parent: 131567
""")
sys.exit(0)
else:
ids=sys.argv[1:]
taxa = taxon.read_nodes()
names,blastname = taxon.read_names()
divs = taxon.read_divisions()
for i in ids:
c=0
if i not in taxa:
sys.stderr.write("Error: ID " + str(i) + " was not found in the taxonomy filei\n")
sys.exit(0)
while taxa[i].parent != '1' and i != '1':
#print " " * c, taxa[i].taxid, "name:", names[i].name, "rank:", taxa[i].rank, "div:", taxa[i].division, "code:", divs[taxa[i].division].code
bn=names[i].name
if i in blastname:
bn=blastname[i].name
print("{} name: {} blast name: {} rank: {} parent: {}".format(taxa[i].taxid, names[i].name, bn, taxa[i].rank, taxa[i].parent))
i=taxa[i].parent
c+=1
import argparse
import taxon
if __name__ == '__main__':
parser = argparse.ArgumentParser(description="Append taxonomy to the patric metadata file. This adds it at column 67")
parser.add_argument('-f', help='patric metadata file', required=True)
parser.add_argument('-o', help='output file', required=True)
parser.add_argument('-t', help='taxonomy directory (default=/home2/db/taxonomy/current/)',
default='/home2/db/taxonomy/current/')
parser.add_argument('-v', help='verbose output', action="store_true")
args = parser.parse_args()
sys.stderr.write("Reading taxonomy\n")
taxa = taxon.read_nodes(directory=args.t)
names, blastname = taxon.read_names(directory=args.t)
divs = taxon.read_divisions(directory=args.t)
sys.stderr.write("Read taxonomy\n")
want = ['superkingdom', 'phylum', 'class', 'order', 'family', 'genus', 'species']
with open(args.o, 'w', encoding='utf-8') as out:
with open(args.f, 'r', encoding='utf-8') as f:
for l in f:
p = l.strip().split("\t")
while (len(p) <= 68):
p.append("")
if l.startswith("genome_id"):
out.write("{}\t{}\n".format(l.strip(), "\t".join(want)))
continue
88770 name: Panarthropoda blast name: Panarthropoda rank: no rank parent: 1206794
1206794 name: Ecdysozoa blast name: Ecdysozoa rank: no rank parent: 33317
33317 name: Protostomia blast name: Protostomia rank: no rank parent: 33213
33213 name: Bilateria blast name: Bilateria rank: no rank parent: 6072
6072 name: Eumetazoa blast name: Eumetazoa rank: no rank parent: 33208
33208 name: Metazoa blast name: animals rank: kingdom parent: 33154
33154 name: Opisthokonta blast name: Opisthokonta rank: no rank parent: 2759
2759 name: Eukaryota blast name: eukaryotes rank: superkingdom parent: 131567
""")
sys.exit(0)
else:
ids = sys.argv[1:]
taxa = taxon.read_nodes()
names, blastname = taxon.read_names()
divs = taxon.read_divisions()
for i in ids:
c = 0
if i not in taxa:
sys.stderr.write("Error: ID " + str(i) +
" was not found in the taxonomy filei\n")
sys.exit(0)
while taxa[i].parent != '1' and i != '1':
#print " " * c, taxa[i].taxid, "name:", names[i].name, "rank:", taxa[i].rank, "div:", taxa[i].division, "code:", divs[taxa[i].division].code
bn = names[i].name
if i in blastname:
bn = blastname[i].name
print(
"{} name: {} blast name: {} rank: {} parent: {}".format(
taxa[i].taxid, names[i].name, bn, taxa[i].rank,
taxa[i].parent))