#!/bin/env python2
import re, sys
from getwiki import GlycanData
from collections import defaultdict
headers = """
accession Hex HexNAc dHex NeuAc NeuGc HexA HexN S P aldi Xxx X Count
""".split()
w = GlycanData()
print "\t".join(headers)
for acc in w.iterglycanid():
g = w.get(acc)
row = defaultdict(lambda: [0, False])
row['accession'] = (acc, False)
for ann in g.annotations(type="MonosaccharideCount", source="EdwardsLab"):
try:
value = [int(ann.get('value')), False]
except ValueError:
value = [int(ann.get('value')[:-1]), True]
prop = ann.get('property')
if prop.endswith('Count'):
prop = prop[:-5]
row[prop] = value
# print row
row['Count'] = row['Monosaccharide']
if 'Xxx' not in row:
row['Xxx'] = [0, False]
for k in row:
for fn in args:
for it in open(fn):
yield it.strip()
gtc = GlyTouCan()
allmotifs = dict()
for acc, label, redend in gtc.allmotifs():
allmotifs[acc] = dict(label=label, redend=redend)
current = set()
for gtcacc in accessions(sys.argv[1:]):
start = time.time()
g = w.get(gtcacc)
newgly = False
if not g:
newgly = True
g = Glycan(accession=gtcacc)
g.delete('wurcs')
g.delete('glycoct')
g.delete('iupac')
g.delete_annotations(source='GlyTouCan', type='Sequence')
g.set_annotation(value=gtc.getseq(gtcacc, 'wurcs'),
property='WURCS',
source='GlyTouCan',
type='Sequence')