def scores(self, id1, id2):
"""Returns the score between two given GO terms.
:arg id1, identifier of a GO term (ie: GO:0043231, or whatever
identifier is in your ontology).
:arg id2, identifier of a GO term (ie: GO:0043229, or whatever
identifier is in your ontology).
"""
golib = PyGoLib(self.goterms)
#golib.fix_go_graph()
goterm1 = self.goterms[id1]
path1 = golib.get_path(goterm1, pred=goterm1['id'], paths=[],
details=True)
#print goterm1['id'], len(path1)
ancester1 = _get_all_ancesters(path1)
semantic_values1 = self.semantic_values(goterm1['id'])
goterm2 = self.goterms[id2]
path2 = golib.get_path(goterm2, pred=goterm2['id'], paths=[],
details=True)
#print goterm2['id'], len(path2)
ancester2 = _get_all_ancesters(path2)
semantic_values2 = self.semantic_values(goterm2['id'])
common_ancester = list(set(ancester1).intersection(set(ancester2)))
sum_comm_anc = 0
for ancester in common_ancester:
sum_comm_anc = sum_comm_anc + semantic_values2[ancester] + \
semantic_values1[ancester]
score = sum_comm_anc / (sum(semantic_values1.values())
+ sum(semantic_values2.values()))
return score
def scores(self, id1, id2):
"""Returns the score between two given GO terms.
:arg id1, identifier of a GO term (ie: GO:0043231, or whatever
identifier is in your ontology).
:arg id2, identifier of a GO term (ie: GO:0043229, or whatever
identifier is in your ontology).
"""
golib = PyGoLib(self.goterms)
goterm1 = self.goterms[id1]
path1 = golib.get_path(goterm1, pred=goterm1['id'], paths=[])
goterm2 = self.goterms[id2]
path2 = golib.get_path(goterm2, pred=goterm2['id'], paths=[])
# We use goterm['id'] instead of the id provided to take into
# account alt_id which are in the list of goterms but not in the
# paths. Via goterm['id'] we get the 'normal' GO term identifier.
scores = self.__score_parents(goterm1['id'], goterm2['id'], path1,
path2)
if scores:
score = min(scores)
self.log.debug("%s and %s are parents" % (id1, id2))
return (score, score)
else:
score = self.__score_cousins(goterm1['id'], goterm2['id'], path1,
path2)
return score
def semantic_values(self, id1):
""" Returns the semantic values of all the parents of a given
term.
:arg id1, identifier of a GO term (ie: GO:0043231, or whatever
identifier is in your ontology).
"""
golib = PyGoLib(self.goterms)
goterm1 = self.goterms[id1]
path1 = golib.get_path(goterm1, pred=goterm1['id'], paths=[],
details=True)
semantic_values = {}
for ancester in _get_all_ancesters(path1):
if ancester == goterm1['id']:
semantic_values[ancester] = 1
continue
tmp_score = []
for item in path1:
tmp_cnt = 1
if ancester in item:
path_el = item.split(',')
ind = path_el.index(ancester)
for step in range(ind - 1, 0, -2):
if path_el[step] == 'is_a':
tmp_cnt = tmp_cnt * 0.8
elif path_el[step] == 'part_of':
tmp_cnt = tmp_cnt * 0.6
tmp_score.append(tmp_cnt)
semantic_values[ancester] = max(tmp_score)
return semantic_values
def semantic_values(self, id1):
""" Returns the semantic values of all the parents of a given
term.
:arg id1, identifier of a GO term (ie: GO:0043231, or whatever
identifier is in your ontology).
"""
golib = PyGoLib(self.goterms)
goterm1 = self.goterms[id1]
path1 = golib.get_path(goterm1,
pred=goterm1['id'],
paths=[],
details=True)
semantic_values = {}
for ancester in _get_all_ancesters(path1):
if ancester == goterm1['id']:
semantic_values[ancester] = 1
continue
tmp_score = []
for item in path1:
tmp_cnt = 1
if ancester in item:
path_el = item.split(',')
ind = path_el.index(ancester)
for step in range(ind - 1, 0, -2):
if path_el[step] == 'is_a':
tmp_cnt = tmp_cnt * 0.8
elif path_el[step] == 'part_of':
tmp_cnt = tmp_cnt * 0.6
tmp_score.append(tmp_cnt)
semantic_values[ancester] = max(tmp_score)
return semantic_values
def scores(self, id1, id2):
"""Returns the score between two given GO terms.
:arg id1, identifier of a GO term (ie: GO:0043231, or whatever
identifier is in your ontology).
:arg id2, identifier of a GO term (ie: GO:0043229, or whatever
identifier is in your ontology).
"""
golib = PyGoLib(self.goterms)
#golib.fix_go_graph()
goterm1 = self.goterms[id1]
path1 = golib.get_path(goterm1,
pred=goterm1['id'],
paths=[],
details=True)
#print goterm1['id'], len(path1)
ancester1 = _get_all_ancesters(path1)
semantic_values1 = self.semantic_values(goterm1['id'])
goterm2 = self.goterms[id2]
path2 = golib.get_path(goterm2,
pred=goterm2['id'],
paths=[],
details=True)
#print goterm2['id'], len(path2)
ancester2 = _get_all_ancesters(path2)
semantic_values2 = self.semantic_values(goterm2['id'])
common_ancester = list(set(ancester1).intersection(set(ancester2)))
sum_comm_anc = 0
for ancester in common_ancester:
sum_comm_anc = sum_comm_anc + semantic_values2[ancester] + \
semantic_values1[ancester]
score = sum_comm_anc / (sum(semantic_values1.values()) +
sum(semantic_values2.values()))
return score
class GoDistanceCounter(object):
""" This class is the main class of the project to compute the
distance between two GO terms.
"""
def __init__(self, data=None):
""" Constructor.
:arg data, the graph of ontologies
"""
self.goterms = data
if self.goterms is None:
self.goterms = {}
self.log = get_logger()
self.pygo = PyGoLib(self.goterms)
def __score_cousins(self, goid1, goid2, path1=None, path2=None):
""" For two given GO term ID and the list of their path, return
the score between them.
:arg goid1, GO term ID (ie: GO:0043231, or whatever identifier is
in your ontology).
:arg goid2, GO term ID (ie: GO:0043229, or whatever identifier is
in your ontology).
:kwarg path1, the list path from the first GO term to the top of
the tree, as returned by get_path().
:kwarg path2, the list path from the second GO term to the top
of the tree, as returned by get_path().
"""
if path1 is None:
path1 = self.pygo.get_path(self.goterms[goid1])
if path2 is None:
path2 = self.pygo.get_path(self.goterms[goid2])
mindist = None
deltalevel = None
for path in path1:
step1 = path.split(',')
for opath in path2:
step2 = opath.split(',')
inter = _get_ancester(step1, step2)
if inter:
index1 = step1.index(inter)
index2 = step2.index(inter)
dist = index1 + index2
deltaleveltmp = abs(index1 - index2)
if not mindist or dist < mindist:
mindist = dist
deltalevel = deltaleveltmp
if mindist is not None and deltalevel is not None:
return (mindist, deltalevel)
else:
return None
def __score_parents(self, goid1, goid2, path1=None, path2=None):
""" For two given GO term ID and the list of their path, return
the score between them if one is parent of the other.
:arg goid1, GO term ID (ie: GO:0043231, or whatever identifier is
in your ontology).
:arg goid2, GO term ID (ie: GO:0043229, or whatever identifier is
in your ontology).
:kwarg path1, the list path from the first GO term to the top of
the tree, as returned by get_path().
:kwarg path2, the list path from the second GO term to the top
of the tree, as returned by get_path().
"""
if path1 is None:
path1 = self.pygo.get_path(self.goterms[goid1])
scores = []
for paths in path1:
steps = paths.split(',')
if goid2 in steps:
start = steps.index(goid1)
stop = steps.index(goid2)
score = abs(stop - start)
scores.append(score)
if path2 is None:
path2 = self.pygo.get_path(self.goterms[goid2])
for paths in path2:
steps = paths.split(',')
if goid1 in steps:
start = steps.index(goid1)
stop = steps.index(goid2)
score = abs(stop - start)
scores.append(score)
return scores
def scores(self, id1, id2):
"""Returns the score between two given GO terms.
:arg id1, identifier of a GO term (ie: GO:0043231, or whatever
identifier is in your ontology).
:arg id2, identifier of a GO term (ie: GO:0043229, or whatever
identifier is in your ontology).
"""
golib = PyGoLib(self.goterms)
goterm1 = self.goterms[id1]
path1 = golib.get_path(goterm1, pred=goterm1['id'], paths=[])
goterm2 = self.goterms[id2]
path2 = golib.get_path(goterm2, pred=goterm2['id'], paths=[])
# We use goterm['id'] instead of the id provided to take into
# account alt_id which are in the list of goterms but not in the
# paths. Via goterm['id'] we get the 'normal' GO term identifier.
scores = self.__score_parents(goterm1['id'], goterm2['id'], path1,
path2)
if scores:
score = min(scores)
self.log.debug("%s and %s are parents" % (id1, id2))
return (score, score)
else:
score = self.__score_cousins(goterm1['id'], goterm2['id'], path1,
path2)
return score