In [21]:

    

fname = '/home/tiago/OTU_GG.txt'
wname = '../analysis_data/old_taxa.txt'


    

In [16]:

    

query_taxa = {}
with open(fname, 'rt') as f:
    l = f.readline()
    while l != '':
        l = l.rstrip()
        if l.startswith('Query='):
            query = l.rstrip().split(' ')[1]
            l = f.readline()
        elif len(l) > 0 and l[0] == '>':
            line = l[1:]
            l = f.readline()            
            while l != '\n':
                
                line += l.rstrip()
                l = f.readline()
            taxon = ' '.join(filter(lambda x: x.find('__') > -1, line.split(' ')))
            query_taxa[query] = taxon
        else:
            l = f.readline()


    

In [23]:

    

with open(wname, 'wt') as w:
    for query, taxon in query_taxa.items():
        w.write('%s\t%s\n' % (query, taxon))