class WrSubObo(object):
"""Read a large GO-DAG from an obo file. Write a subset GO-DAG into a small obo file."""
def __init__(self, fin_obo):
self.fin_obo = fin_obo
self.godag = GODag(fin_obo)
def wrobo(self, fout_obo, goid_sources):
"""Write a subset obo file containing GO ID sources and their parents."""
goids_all = self._get_all_parents(goid_sources).union(goid_sources)
b_trm = False
b_prt = True
with open(fout_obo, 'w') as prt:
self._prt_info(prt, goid_sources, goids_all)
with open(self.fin_obo) as ifstrm:
for line in ifstrm:
if not b_trm:
if line[:6] == "[Term]":
b_trm = True
b_prt = False
else:
if line[:6] == 'id: GO':
b_trm = False
b_prt = line[4:14] in goids_all
if b_prt:
prt.write("[Term]\n")
if b_prt:
prt.write(line)
sys.stdout.write(" WROTE {N} GO TERMS: {OBO}\n".format(
N=len(goids_all), OBO=fout_obo))
def _prt_info(self, prt, goid_sources, goids_all):
"""Print information describing how this obo setset was created."""
prt.write(
"Contains {N} GO IDs. Created using {M} GO sources:\n".format(
N=len(goids_all), M=len(goid_sources)))
for goid in goid_sources:
prt.write(" {GO}\n".format(GO=str(self.godag.get(goid, ""))))
prt.write("\n")
def _get_all_parents(self, goid_sources):
"""Get all GO ID parents for all GO IG sources."""
parents = set()
for goid in goid_sources:
goterm = self.godag.get(goid, None)
if goterm is not None:
parents |= goterm.get_all_parents()
return parents
class WrSubObo(object):
"""Read a large GO-DAG from an obo file. Write a subset GO-DAG into a small obo file."""
def __init__(self, fin_obo, optional_attrs=None, load_obsolete=None):
self.fin_obo = fin_obo
self.godag = GODag(fin_obo, optional_attrs, load_obsolete)
self.relationships = optional_attrs is not None and 'relationship' in optional_attrs
def wrobo(self, fout_obo, goid_sources):
"""Write a subset obo file containing GO ID sources and their parents."""
goids_all = self._get_goids_all(goid_sources)
b_trm = False
b_prt = True
with open(fout_obo, 'w') as prt:
self._prt_info(prt, goid_sources, goids_all)
with open(self.fin_obo) as ifstrm:
for line in ifstrm:
if not b_trm:
if line[:6] == "[Term]":
b_trm = True
b_prt = False
elif line[:6] == "[Typedef]":
b_prt = True
else:
if line[:6] == 'id: GO':
b_trm = False
b_prt = line[4:14] in goids_all
if b_prt:
prt.write("[Term]\n")
if b_prt:
prt.write(line)
sys.stdout.write(" WROTE {N} GO TERMS: {OBO}\n".format(
N=len(goids_all), OBO=fout_obo))
def _get_goids_all(self, go_sources):
"""Given GO ID sources and optionally the relationship attribute, return all GO IDs."""
go2obj_user = {}
objrel = CurNHigher(self.relationships, self.godag)
objrel.get_go2obj_cur_n_high(go2obj_user, go_sources)
goids = set(go2obj_user)
for goterm in go2obj_user.values():
if goterm.alt_ids:
goids.update(goterm.alt_ids)
return goids
def _prt_info(self, prt, goid_sources, goids_all):
"""Print information describing how this obo setset was created."""
prt.write(
"! Contains {N} GO IDs. Created using {M} GO sources:\n".format(
N=len(goids_all), M=len(goid_sources)))
for goid in goid_sources:
prt.write("! {GO}\n".format(GO=str(self.godag.get(goid, ""))))
prt.write("\n")
class WrSubObo(object):
"""Read a large GO-DAG from an obo file. Write a subset GO-DAG into a small obo file."""
def __init__(self, fin_obo=None, optional_attrs=None, load_obsolete=None):
self.fin_obo = fin_obo
self.godag = GODag(fin_obo, optional_attrs,
load_obsolete) if fin_obo is not None else None
self.relationships = optional_attrs is not None and 'relationship' in optional_attrs
def wrobo(self, fout_obo, goid_sources):
"""Write a subset obo file containing GO ID sources and their parents."""
goids_all = self._get_goids_all(goid_sources)
with open(fout_obo, 'w') as prt:
self._prt_info(prt, goid_sources, goids_all)
self.prt_goterms(prt, self.fin_obo, goids_all)
print(" WROTE {N} GO TERMS: {OBO}\n".format(N=len(goids_all),
OBO=fout_obo))
@staticmethod
def prt_goterms(fin_obo, goids, prt, b_prt=True):
"""Print the specified GO terms for GO IDs in arg."""
b_trm = False
with open(fin_obo) as ifstrm:
for line in ifstrm:
if not b_trm:
if line[:6] == "[Term]":
b_trm = True
b_prt = False
elif line[:6] == "[Typedef]":
b_prt = True
else:
if line[:6] == 'id: GO':
b_trm = False
b_prt = line[4:14] in goids
if b_prt:
prt.write("[Term]\n")
if b_prt:
prt.write(line)
@staticmethod
def get_goids(fin_obo, name):
"""Get GO IDs whose name matches given name."""
goids = set()
# pylint: disable=unsubscriptable-object
goterm = None
with open(fin_obo) as ifstrm:
for line in ifstrm:
if goterm is not None:
semi = line.find(':')
if semi != -1:
goterm[line[:semi]] = line[semi + 2:].rstrip()
else:
if name in goterm['name']:
goids.add(goterm['id'])
goterm = None
elif line[:6] == "[Term]":
goterm = {}
return goids
def _get_goids_all(self, go_sources):
"""Given GO ID sources and optionally the relationship attribute, return all GO IDs."""
go2obj_user = {}
objrel = CurNHigher(self.relationships, self.godag)
objrel.get_go2obj_cur_n_high(go2obj_user, go_sources)
goids = set(go2obj_user)
for goterm in go2obj_user.values():
if goterm.alt_ids:
goids.update(goterm.alt_ids)
return goids
def _prt_info(self, prt, goid_sources, goids_all):
"""Print information describing how this obo setset was created."""
prt.write(
"! Contains {N} GO IDs. Created using {M} GO sources:\n".format(
N=len(goids_all), M=len(goid_sources)))
for goid in goid_sources:
prt.write("! {GO}\n".format(GO=str(self.godag.get(goid, ""))))
prt.write("\n")
from goatools.obo_parser import GODag
obodag = GODag("go-basic.obo")
for i,line in enumerate(open('new_annotations.csv')):
if i > 0:
la = line.split("\t")
gos = la[1].split(',')
print(la[0])
outline = []
for go in gos:
if go.startswith("GO:"):
outline.append("{} ({})".format(
obodag.get(go).name,
go))
print(",".join(outline))
class WrSubObo(object):
"""Read a large GO-DAG from an obo file. Write a subset GO-DAG into a small obo file."""
def __init__(self, fin_obo=None, optional_attrs=None, load_obsolete=None):
self.fin_obo = fin_obo
self.godag = GODag(fin_obo, optional_attrs, load_obsolete) if fin_obo is not None else None
self.relationships = optional_attrs is not None and 'relationship' in optional_attrs
def wrobo(self, fout_obo, goid_sources):
"""Write a subset obo file containing GO ID sources and their parents."""
goids_all = self._get_goids_all(goid_sources)
with open(fout_obo, 'w') as prt:
self._prt_info(prt, goid_sources, goids_all)
self.prt_goterms(prt, self.fin_obo, goids_all)
print(" WROTE {N} GO TERMS: {OBO}\n".format(N=len(goids_all), OBO=fout_obo))
@staticmethod
def prt_goterms(fin_obo, goids, prt, b_prt=True):
"""Print the specified GO terms for GO IDs in arg."""
b_trm = False
with open(fin_obo) as ifstrm:
for line in ifstrm:
if not b_trm:
if line[:6] == "[Term]":
b_trm = True
b_prt = False
elif line[:6] == "[Typedef]":
b_prt = True
else:
if line[:6] == 'id: GO':
b_trm = False
b_prt = line[4:14] in goids
if b_prt:
prt.write("[Term]\n")
if b_prt:
prt.write(line)
@staticmethod
def get_goids(fin_obo, name):
"""Get GO IDs whose name matches given name."""
goids = set()
# pylint: disable=unsubscriptable-object
goterm = None
with open(fin_obo) as ifstrm:
for line in ifstrm:
if goterm is not None:
semi = line.find(':')
if semi != -1:
goterm[line[:semi]] = line[semi+2:].rstrip()
else:
if name in goterm['name']:
goids.add(goterm['id'])
goterm = None
elif line[:6] == "[Term]":
goterm = {}
return goids
def _get_goids_all(self, go_sources):
"""Given GO ID sources and optionally the relationship attribute, return all GO IDs."""
go2obj_user = {}
objrel = CurNHigher(self.relationships, self.godag)
objrel.get_id2obj_cur_n_high(go2obj_user, go_sources)
goids = set(go2obj_user)
for goterm in go2obj_user.values():
if goterm.alt_ids:
goids.update(goterm.alt_ids)
return goids
def _prt_info(self, prt, goid_sources, goids_all):
"""Print information describing how this obo setset was created."""
prt.write("! Contains {N} GO IDs. Created using {M} GO sources:\n".format(
N=len(goids_all), M=len(goid_sources)))
for goid in goid_sources:
prt.write("! {GO}\n".format(GO=str(self.godag.get(goid, ""))))
prt.write("\n")