def map(attr, predicate):
cell = getattr(self, attr)()
value = cell.value
if value:
for iri, label in list(self.sheet.convert_cell(value, predicate=predicate)):
if ',' not in value and label != value:
log.warning(f'label mismatch {label!r} != {value!r}')
if iri is None:
if label == 'bed nucleus of stria terminalis juxtacapsular nucleus':
iri = OntTerm('UBERON:0011173', label='anterior division of bed nuclei of stria terminalis')
else:
log.debug(f'nothing found for {label}')
continue
if isinstance(iri, tuple):
op, *rest = iri # TODO need combinators in future version for union/intersection of object
out = (op, *(NegPhenotype(r, predicate)
if isinstance(r, LacksObject) else
Phenotype(r, predicate)
for r in rest if r is not None))
yield out
continue
elif isinstance(iri, LacksObject):
p = NegPhenotype(iri.asURIRef(), predicate)
else:
p = Phenotype(iri, predicate)
yield p.asIndicator()
def mapCell(cls, cell, syns=False):
search_prefixes = (
'UBERON',
'CHEBI',
'PR',
'NCBITaxon',
'NCBIGene',
'ilxtr',
'NIFEXT',
'SAO',
'NLXMOL',
'BIRNLEX',
)
if ':' in cell and ' ' not in cell:
log.debug(cell)
if 'http' in cell:
if cell.startswith('http'):
t = OntTerm(iri=cell)
else:
return None, None # garbage with http inline
else:
t = OntTerm(cell, exclude_prefix=(
'FMA', )) # FIXME need better error message in ontquery
return t.u, t.label
result = [
r for r in cls.sgv.findByTerm(
cell, searchSynonyms=syns, prefix=search_prefixes)
if not r['deprecated']
]
#printD(cell, result)
if not result:
log.debug(f'{cell}')
maybe = list(cls.query(label=cell, exclude_prefix=('FMA', )))
if maybe:
t = maybe[0]
return t.u, t.label
elif not syns:
return cls.mapCell(cell, syns=True)
else:
return None, None
elif len(result) > 1:
#printD('WARNING', result)
result = select_by_curie_rank(result)
else:
result = result[0]
return rdflib.URIRef(result['iri']), result['labels'][0]
def neuron_existing(self):
curie = self.curie().value
if curie:
id_ = OntTerm(curie).u
match = self.sheet.existing.get(id_)
if match:
return match
al = self.alignment_label().value
return self.sheet.existing.get(al if al else self.label().value)
def neuron_existing(self):
curie = self.curie().value
if curie:
id_ = OntTerm(curie).u
match = self.sheet.existing.get(id_)
if match:
match.id_ = id_
return match
al = self.alignment_label().value
nrn = self.sheet.existing.get(al if al else self.label().value)
if nrn and curie:
nrn.id_ = id_
return nrn
def entailed_molecular_phenotypes(self):
cell = self.exhasmolecularphenotype()
labels = cell.value.split(',')
# FIXME hack
yield OntTerm(curie='ilxtr:GABAReceptor').asPhenotype()
yield OntTerm(curie='ilxtr:glutamateReceptor').asPhenotype()
for label in labels:
label = label.strip()
term = self.sheet.sgv.findByTerm(label)
if term:
ot = OntTerm(iri=term[0]['iri'])
p = ot.asPhenotype()
yield p
if ot.curie == 'NLXMOL:1006001':
yield OntTerm(curie='ilxtr:GABAReceptor').asPhenotype()
elif ot.curie == 'SAO:1164727693':
yield OntTerm(curie='ilxtr:glutamateReceptor').asPhenotype()
def mapCell(cls, cell, syns=False, predicate=None):
search_prefixes = ('UBERON', 'CHEBI', 'PR', 'NCBIGene', 'NCBITaxon',
'ilxtr', 'NIFEXT', 'SAO', 'NLXMOL', 'BIRNLEX',)
if predicate and predicate in Phenotype._molecular_predicates:
# uberon syns pollute molecular results so move it to one before birnlex
ub, *rest, b = search_prefixes
search_prefixes = (*rest, ub, b)
if cell == 'contralateral':
return ilxtr.Contralateral, cell # XXX FIXME only BSPO has this right now
elif cell.lower() == 'gaba receptor role':
return ilxtr.GABAReceptor, cell
if ':' in cell and ' ' not in cell:
log.debug(cell)
if 'http' in cell:
if cell.startswith('http'):
t = OntTerm(iri=cell)
else:
return None, None # garbage with http inline
else:
t = OntTerm(cell, exclude_prefix=('FMA',)) # FIXME need better error message in ontquery
return t.u, t.label
if cell in ('Vertebrata', ): # search syns
syns = True
def rank_mask(r):
"""
create a boolean array testing if the current entry
starts with the prefixes in order and what you will
get out is arrays where the nth element is true if
the nth prefix is matched which will then be sorted by n
1 0 0 0 0 0 0 1 \\
1 0 0 0 0 0 0 0 \\
0 1 0 0 0 0 0 0 \\
0 0 1 0 0 0 0 0 \\
0 0 0 1 0 0 0 0 \\
0 0 0 0 1 0 0 0 \\
"""
# why did it take so long to think of this?
return (
*(r['curie'].startswith(p) for p in search_prefixes),
'labels' in r and cell in r['labels'],
)
result = sorted([r for r in cls.sgv.findByTerm(cell, searchSynonyms=syns, prefix=search_prefixes)
if not r['deprecated']], key=rank_mask, reverse=True)
#printD(cell, result)
if not result:
log.debug(f'{cell}')
maybe = list(cls.query(label=cell, exclude_prefix=('FMA',)))
if maybe:
t = maybe[0]
return t.u, t.label
elif not syns:
return cls.mapCell(cell, syns=True, predicate=predicate)
else:
return None, None
elif len(result) > 1:
#printD('WARNING', result)
result = result[0] #select_by_curie_rank(result)
else:
result = result[0]
return rdflib.URIRef(result['iri']), result['labels'][0]
def loop_internal(j, header, cell):
nonlocal id
nonlocal current_neuron
nonlocal do_release
notes = list(process_note(get_note(i + 1, j, self.cells_index))) # + 1 since headers is removed
if notes and not header.startswith('has'):
_predicate = self.convert_other(header)
if cell:
_object = rdflib.Literal(cell) # FIXME curies etc.
else:
_object = rdf.nil
other_notes[_predicate, _object] = notes
if header == 'curie':
id = OntId(cell).u if cell else None
return
elif header == 'label':
if id == OntId('NIFEXT:66').u:
breakpoint()
label_neuron = cell
if cell in self.existing:
current_neuron = self.existing[cell]
elif cell:
# TODO
self.new.append(cell)
else:
raise ValueError(cell) # wat
return
elif header == 'Status':
# TODO
if cell == 'Yes':
do_release = True
elif cell == 'Maybe':
pass
elif cell == 'Not yet':
pass
elif cell == 'Delete':
pass
else:
pass
return
elif header == 'PMID':
# TODO
return
elif header == 'Other reference':
# TODO
return
elif header == 'Other label':
# TODO
return
elif header == 'definition':
return # FIXME single space differences between the spreadsheet and the source
if cell:
definition_neuron = rdflib.Literal(cell)
elif header == 'synonyms':
if cell:
synonyms_neuron = [rdflib.Literal(s.strip())
# FIXME bare comma is extremely dangerous
for s in cell.split(',')]
return
elif header in self.skip:
return
objects = []
if cell:
predicate = self.convert_header(header)
if predicate is None:
log.debug(f'{(header, cell, notes)}')
for object, label in self.convert_cell(cell):
if predicate in NeuronCUT._molecular_predicates:
if isinstance(object, tuple):
op, *rest = object
rest = [OntTerm(o).asIndicator().URIRef for o in rest]
object = op, *rest
elif object:
log.debug(f'{object!r}')
object = OntTerm(object).asIndicator().URIRef
if isinstance(label, tuple): # LogicalPhenotype case
_err = []
for l in label:
if self.lower_check(l, cell):
_err.append((cell, label))
if _err:
self.errors.extend(_err)
else:
objects.append(object)
elif self.lower_check(label, cell):
self.errors.append((cell, label))
elif str(id) == object:
self.errors.append((header, cell, object, label))
object = None
else:
objects.append(object)
if notes:
# FIXME this is a hack to only attach to the last value
# since we can't distinguish at the moment
wat[predicate, object] = notes
if object is not None:
# object aka iri can be none if we don't find anything
object_notes[object] = notes
else:
predicate_notes[predicate] = notes
# FIXME it might also be simpler in some cases
# to have this be object_notes[object] = notes
# because we are much less likely to have the same
# phenotype appear attached to the different dimensions
# FIXME comma sep is weak here because the
# reference is technically ambiguous
# might be an argument for the denormalized form ...
# or perhaps having another sheet for cases like that
else:
return
if predicate and objects:
for object in objects: # FIXME has layer location phenotype
if isinstance(object, tuple):
op, *rest = object
pes = (Phenotype(r, predicate) for r in rest) # FIXME nonhomogenous phenotypes
phenotypes.append(LogicalPhenotype(op, *pes))
elif object:
phenotypes.append(Phenotype(object, predicate))
else:
self.errors.append((object, predicate, cell))
elif objects:
self.errors.append((header, objects))
else:
self.errors.append((header, cell))