make_blobolog_file.py

gistfile1.py

import sys
from collections import defaultdict
from itertools import izip_longest

def grouper(n, iterable, fillvalue=None):
    args = [iter(iterable)] * n
    return izip_longest(fillvalue=fillvalue, *args)

contigs = defaultdict(dict)

for ln in open(sys.argv[1]):
        cols = ln.rstrip().split("\t")
        if len(cols) > 1:
                taxon = cols[0]
                for level, tax, ignore in grouper(3, cols[1:]):
                        contigs[taxon][level] = tax

for ln in open(sys.argv[2]):
        cols = ln.rstrip().split("\t")
        if cols[1] in contigs and sys.argv[3] in contigs[cols[1]]:
                cols.append(contigs[cols[1]][sys.argv[3]])
        else:
                cols.append("Not annotated")
        print "\t".join(cols)