def accs():
global gtc
if len(sys.argv) > 1:
gtc = GlyTouCan(prefetch=False, usecache=False)
for acc in sys.argv[1:]:
yield acc
else:
gtc = GlyTouCan(prefetch=True, usecache=False)
for m in w.itermotif():
yield m.get('id')
def __init__(self, **kwargs):
assert kwargs.get('accession') != None
kwargs['collection'] = self.id
if kwargs.get('glytoucan') == None:
kwargs['glytoucan'] = kwargs['accession']
if kwargs.get('wurcs') == None or kwargs.get('glycoct') == None:
if not self.gtc:
self.gtc = GlyTouCan()
if kwargs.get('wurcs') == None:
kwargs['wurcs'] = self.gtc.getseq(kwargs['glytoucan'], 'wurcs')
if kwargs.get('glycoct') == None:
kwargs['glycoct'] = self.gtc.getseq(kwargs['glytoucan'],
'glycoct')
super(GlyTouCanMotif, self).__init__(**kwargs)
class GlyTouCanMotif(Motif):
gtc = None
id = 'GTC'
def __init__(self, **kwargs):
assert kwargs.get('accession') != None
kwargs['collection'] = self.id
if kwargs.get('glytoucan') == None:
kwargs['glytoucan'] = kwargs['accession']
if kwargs.get('wurcs') == None or kwargs.get('glycoct') == None:
if not self.gtc:
self.gtc = GlyTouCan(usecache=False, prefetch=False)
if kwargs.get('wurcs') == None:
kwargs['wurcs'] = self.gtc.getseq(kwargs['glytoucan'], 'wurcs')
if kwargs.get('glycoct') == None:
kwargs['glycoct'] = self.gtc.getseq(kwargs['glytoucan'],
'glycoct')
super(GlyTouCanMotif, self).__init__(**kwargs)
def accessions():
global gtc
if len(sys.argv) == 2 and sys.argv[1] == "-":
gtc = GlyTouCan(usecache=False)
for acc in sys.stdin:
yield acc.strip()
elif len(sys.argv) == 2 and sys.argv[1] == "*":
gtc = GlyTouCan(usecache=False)
for acc in gtc.allaccessions():
if acc in replace:
continue
yield acc
else:
gtc = GlyTouCan(prefetch=False)
for acc in sys.argv[1:]:
yield acc
import findpygly
from pygly.GlycanResource import GlyTouCan
def accessions(args):
if len(args) == 0:
for it in sys.stdin:
yield it.strip()
else:
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)
#!/bin/env python2
import sys, glob
import findpygly
from pygly.GlycanResource import GlyTouCan, GlyCosmos
seqtype = sys.argv[1].split('/')[-1]
assert (seqtype in ('wurcs', 'glycoct', 'genglycoct'))
allfn = set(glob.glob(sys.argv[1] + "/G*.txt"))
gco = GlyCosmos(usecache=False)
archived = set(map(lambda d: d['accession'], gco.archived()))
gtc = GlyTouCan(verbose=False, usecache=False)
for acc in gtc.allaccessions():
if acc in archived:
continue
filename = sys.argv[1] + "/" + acc + ".txt"
if filename in allfn:
allfn.remove(filename)
continue
seq = None
if seqtype == 'wurcs':
seq = gtc.getseq(acc, 'wurcs')
if seqtype == 'glycoct':
#!/bin/env python2
import findpygly
# from pygly.GlyTouCan import GlyTouCan
from pygly.GlycanResource import GlyTouCan as GTC
import sys, re, time
import urllib
import hashlib
gtc = GTC(prefetch=False)
# Need wurcs skeleton codes for UniCarbKB and Byonic symbols...
symbol2wurcs_definition = """
NeuAc AUd21122h_5*NCC/3=O 1
NeuGc AUd21122h_5*NCCO/3=O 2
Fuc u1221m 6
Hex uxxxxh 5
HexNAc uxxxxh_2*NCC/3=O 4
dHex uxxxxm 3
Pent uxxxh 10
P *OPO/3O/3=O -1
Phospho *OPO/3O/3=O -1
S *OSO/3=O/3=O -1
Sulpho *OSO/3=O/3=O -1
"""
symbol2wurcs = {}
wurcsorder = {}
for l in symbol2wurcs_definition.splitlines():
if not l.strip():
continue
for it in open(fn):
yield it.strip()
notation = 'snfg'
style = 'extended'
format = 'svg'
if len(sys.argv) > 1:
notation = sys.argv[1]
if len(sys.argv) > 2:
style = sys.argv[2]
if len(sys.argv) > 3:
format = sys.argv[3]
gtc = GlyTouCan(usecache=False)
for gtcacc in accessions(sys.argv[4:]):
imgfn = "%s.%s" % (gtcacc, format)
if os.path.exists(imgfn):
continue
imgstr = gtc.getimage(gtcacc,
style=style,
notation=notation,
format=format)
if not imgstr:
if style == "extended":
try:
imgstr = urllib.urlopen(
"https://image.glycosmos.org/%s/%s/%s" % (
notation,
format,
#
# G45924NL is floating in-link substituent - IMO this is not a correct representation
# G99993XU ditto
# G94401PZ ditto
# G10633KT ditto
# G48302UE ditto
#
badacc = set("""
""".split())
# f = open('../data/basecomplist1.txt','w')
# Make sure we get the latest version of everything
gtc = GlyTouCan(usecache=False)
def accessions():
if len(sys.argv) == 2 and sys.argv[1] == "-":
for acc in sys.stdin:
yield acc.strip()
elif len(sys.argv) == 2 and sys.argv[1] == "*":
for acc in gtc.allaccessions():
yield acc
else:
for acc in sys.argv[1:]:
yield acc
allskel = set()
seen = defaultdict(set)
#!/bin/env python27
import sys
from getwiki import GlycoMotifWiki, GlyTouCanMotif
w = GlycoMotifWiki()
import findpygly
from pygly.GlycanResource import GlyTouCan
gtc = GlyTouCan(usecache=False)
current = set()
for m, l, re in sorted(gtc.allmotifs()):
motif = GlyTouCanMotif(accession=m,
prefname=l,
name=l,
redend=re,
wurcs=gtccache.gtc2wurcs(m),
glycoct=gtccache.gtc2glycoct(m))
if w.update(motif):
print >> sys.stderr, m
current.add(m)
for m in w.itermotif(collection=GlyTouCanMotif):
if m.get('accession') not in current:
print >> sys.stderr, "Deleting:", m.get('pagename')
w.delete(m.get('pagename'))
import findpygly
from pygly.GlycanResource import GlyTouCan
from pygly.GlycanResource import GlyCosmos
def accessions(args):
if len(args) == 0:
for it in sys.stdin:
yield it.strip()
else:
for fn in args:
for it in open(fn):
yield it.strip()
gtc = GlyTouCan(verbose=False, usecache=False)
gco = GlyCosmos(verbose=False, usecache=False)
allgco = set(gco.allaccessions())
# allmotifs = dict()
# for acc,label,redend in gtc.allmotifs():
# allmotifs[acc] = dict(label=label,redend=redend)
archived = set(map(lambda d: d['accession'], gco.archived()))
print "%d accessions archived." % (len(archived), )
current = set()
for gtcacc in accessions(sys.argv[1:]):
start = time.time()