An online fungal foray

fungi_step1.py

from Bio import Entrez

#Let NCBI know who you are in case you do something stupid :)
Entrez.email = '[email protected]'

search_s ='"ectomycorrhizal root tip" AND "New Zealand"'
handle = Entrez.esearch(db='nucleotide', term=search_s, retmax=100)
ids = Entrez.read(handle)['IdList']

ids[:5]
#gives us ['157086858', '157086857', '157086856', '157086855', '157086854']

fungi_step2.py

from Bio import SeqIO

rec_handle = Entrez.efetch(db='nuccore', rettype='gb', id = ', '.join(ids))
recs = SeqIO.parse(rec_handle, 'gb')

fungi_step3.py

from collections import defaultdict

host_dict = defaultdict(list)

for record in recs:
  for feature in record.features:
    if feature.type == 'source':
      try:  
        host = feature.qualifiers['host'][0]
        taxonomy = ':'.join(record.annotations['taxonomy'])
        host_dict[host].append(taxonomy)
        #can only be one host, so don't need to look other features
        break
      except KeyError:
        #no 'host' in source, will have to dig into these manually
        print 'No host for {0}!'.format(record.id)
        

fungi_step4.py

#they're all silver beech! Let's count unique taxonomies for that species

frequencies = defaultdict(int)

for tax_string in host_dict['Nothofagus menziesii']:
  frequencies[tax_string] += 1

frequencies.values()
#gives us:
#[15, 5, 24, 6, 3, 28, 2]