class RfamFamily(object):
id = attr.ib(validator=is_a(str))
name = attr.ib(validator=is_a(str))
so_terms = attr.ib(validator=is_a(set))
rna_type = attr.ib(validator=is_a(tuple))
@classmethod
def build_all(cls, link_file, family_file):
so_terms = coll.defaultdict(set)
with open(link_file, 'r', 'utf-8') as raw:
for line in raw:
parts = line.split()
if parts[1] != 'SO':
continue
so_terms[parts[0]].add('SO:%s' % parts[2])
families = []
with open(family_file, 'r', 'iso-8859-1') as raw:
for row in csv.reader(raw, delimiter='\t'):
family = row[0]
name = row[1]
rna_type = row[18]
families.append(cls(
id=family,
name=name,
so_terms=so_terms[family],
rna_type=rna_type_to_key(rna_type)
))
return families
class WithFallBacks(object):
from_manual = attr.ib(validator=is_a(ManualInference))
from_name = attr.ib(validator=is_a(FromName))
from_rna_type = attr.ib(validator=is_a(FromRnaType))
from_so_terms = attr.ib(validator=is_a(FromSoTerms))
so_term_search = attr.ib(validator=is_a(SoTermSearch))
@classmethod
def build(cls, manual_file, obo_file, max_depth):
return cls(
from_manual=ManualInference.build(manual_file),
from_name=FromName.build(manual_file),
from_rna_type=FromRnaType.build(manual_file),
from_so_terms=FromSoTerms.build(manual_file),
so_term_search=SoTermSearch.build(manual_file, obo_file, max_depth),
)
@property
def name(self):
return 'fallbacks'
def simplify(self, result):
if not result:
return result
# Remove misc_RNA if possible.
if len(result) > 1 and INSDCTypes.misc_RNA in result:
result = result.remove(INSDCTypes.misc_RNA)
# Remove other if possible. We remove misc_RNA first because other is
# more specific.
if len(result) > 1 and INSDCTypes.other in result:
result = result.remove(INSDCTypes.other)
return result
def __call__(self, family):
result = self.from_manual(family) or \
self.from_name(family) or \
self.from_so_terms(family) or \
self.from_rna_type(family)
if not result:
possible = self.so_term_search(family)
if possible and possible.rna_types != {INSDCTypes.other}:
result = possible
if not result:
return InferredRfamType(
family=family,
method=self.name,
rna_types=frozenset()
)
return self.simplify(result)
class InferredRfamType(object):
family = attr.ib(validator=is_a(RfamFamily))
method = attr.ib(validator=is_a(str))
rna_types = attr.ib(validator=is_a(frozenset))
@classmethod
def build(cls, family, name, result):
rna_types = set()
if isinstance(result, str):
rna_types.add(result)
elif isinstance(result, (list, set, tuple)):
rna_types.update(result)
elif result is None:
pass
else:
raise ValueError("Unknown type of result")
final = set()
for rna_type in rna_types:
if rna_type == 'antisense':
rna_type = 'antisense_RNA'
if rna_type is None:
continue
final.add(getattr(INSDCTypes, rna_type))
return cls(
family=family,
method=name,
rna_types=frozenset(final),
)
def remove(self, value):
if value not in self.rna_types:
return self
return attr.assoc(
self,
rna_types=frozenset(r for r in self.rna_types if r != value)
)
def simple(self):
return {
'family': self.family.id,
'method': self.method,
'rna_types': ';'.join(r.name for r in self.rna_types),
}
def __contains__(self, value):
return value in self.rna_types
def __len__(self):
return len(self.rna_types)
def __bool__(self):
return bool(self.rna_types)
class SoTermSearch(object):
graph = attr.ib()
max_depth = attr.ib(validator=is_a(int))
@classmethod
def build(cls, manual_file, filename, max_depth):
with open(manual_file, 'r', 'utf-8') as handle:
loaded = json.load(handle)
assignments = loaded['assignments']
graph = read_obo(filename)
for so_term, isndc in assignments.items():
graph.node[so_term]['isndc'] = isndc
return cls(
graph=graph,
max_depth=max_depth
)
@property
def name(self):
return 'so-search'
def dfs(self, term, depth):
if term not in self.graph:
return set()
node = self.graph.node[term]
if not depth and 'isndc' in node:
return set([node['isndc']])
if depth:
found = set()
edges = self.graph.out_edges_iter(term, data=True)
for (_, child, data) in edges:
found.update(self.dfs(child, depth - 1))
return found
return set()
def search(self, root):
for depth in range(0, self.max_depth):
found = self.dfs(root, depth)
if found:
return found
return set()
def __call__(self, family):
rna_types = set()
for so_term in family.so_terms:
rna_types.update(self.search(so_term))
return InferredRfamType.build(family, self.name, rna_types)
class ManualInference(object):
assignments = attr.ib(validator=is_a(dict))
@classmethod
def build(cls, filename):
with open(filename, 'r', 'utf-8') as handle:
loaded = json.load(handle)
return cls(assignments=loaded['hardcoded'])
@property
def name(self):
return 'manual'
def __call__(self, family):
return InferredRfamType.build(
family,
self.name,
self.assignments.get(family.id, None)
)
class FromSoTerms(object):
mapping = attr.ib(validator=is_a(dict))
@classmethod
def build(cls, manual_file):
with open(manual_file, 'r', 'utf-8') as handle:
loaded = json.load(handle)
return cls(mapping=loaded['assignments'])
@property
def name(self):
return 'so-term'
def __call__(self, family):
mapped = set(self.mapping.get(so, None) for so in family.so_terms)
return InferredRfamType.build(
family,
self.name,
mapped
)
class FromName(object):
informative_names = attr.ib(validator=is_a(dict))
@classmethod
def build(cls, filename):
with open(filename, 'r', 'utf-8') as handle:
loaded = json.load(handle)
return cls(
informative_names=loaded['informative_names'],
)
@property
def name(self):
return 'name'
def __call__(self, family):
for pattern, rna_type in self.informative_names.items():
if re.search(pattern, family.name, re.IGNORECASE):
return InferredRfamType.build(family, self.name, rna_type)
return InferredRfamType.build(family, self.name, None)
class FromRnaType(object):
mapping = attr.ib(validator=is_a(dict))
@classmethod
def build(cls, filename):
with open(filename, 'r', 'utf-8') as handle:
loaded = json.load(handle)
given = loaded['rna_type_mapping']
return cls(
mapping={rna_type_to_key(r): v for r, v in given.items()}
)
@property
def name(self):
return 'rna-type'
def __call__(self, family):
return InferredRfamType.build(
family,
self.name,
self.mapping.get(family.rna_type, None)
)
class RfamMatchStatus(object):
"""
This represents implied problems from a match between an Rfam family and an
Rna sequence. Problems are detected by various objects and this simply
records which ones have found issues as well as some data about the issues.
This serves as a simple way to organize many possible issues that could be
detected.
"""
has_issue = attr.ib(validator=is_a(bool))
upi = attr.ib(validator=is_a(six.string_types))
taxid = attr.ib()
finders = attr.ib(validator=is_a(list))
messages = attr.ib(validator=is_a(list))
@classmethod
def with_issue(cls, upi, taxid, finder, msg):
"""
Create a new instance that indicates that the given finder has found an
issue specified in the given message.
"""
return cls(has_issue=True,
upi=upi,
taxid=taxid,
finders=[finder],
messages=[msg])
@classmethod
def no_issues(cls, upi, taxid):
"""
Create a new instance that indicates there are no issues.
"""
return cls(has_issue=False,
upi=upi,
taxid=taxid,
finders=[],
messages=[])
@property
def names(self):
"""
Get the names of all finders that have found issues.
"""
return sorted([finder.name for finder in self.finders])
def merge(self, status):
"""
Merge the given status with this one. This will update the issues found
if any.
"""
if status.upi != self.upi and self.taxid == status.taxid:
raise ValueError("Can only merge MatchStatus from the same RNA.")
self.finders.extend(status.finders)
self.messages.extend(status.messages)
self.has_issue = (self.has_issue or status.has_issue)
return self
def as_simple_data(self):
"""
Create a simplified dict representation of this data. This is useful
for storage.
"""
return {
'has_issue': self.has_issue,
'problems': [{
'name': n
} for n in self.names],
}
def as_json(self):
"""
Create a JSON representation of the simplified data.
"""
return json.dumps(self.as_simple_data())
