class SwansonLabels(ParcOnt): # FIXME not labels...
filename = 'swanson'
name = 'Swanson 2014 partonomies'
shortname = 'swannt'
imports = parcCore,
prefixes = {**makePrefixes('NIFRID', 'ilxtr', 'prov'),
'swanr':interlex_namespace('swanson/uris/readable/'),
'SWAN':interlex_namespace('swanson/uris/neuroanatomical-terminology/terms/'),
'SWAA':interlex_namespace('swanson/uris/neuroanatomical-terminology/appendix/'),}
sources = SwansonAppendix,
namespace = prefixes['SWAN']
root = LabelRoot(iri=nsExact(namespace), # FIXME this is not really a label in the strict sense
label='Swanson label root',
shortname=shortname,
definingArtifacts=(s.artifact.iri for s in sources),)
def _triples(self):
for s, p, o in swanson().g:
#if p != rdf.type and o != owl.Ontology:
if s != rdflib.URIRef('http://ontology.neuinfo.org/NIF/ttl/generated/swanson_hierarchies.ttl'):
if p == rdfs.subClassOf and o == ilxtr.swansonBrainRegionConcept:
yield s, p, self.root.iri
elif p == rdfs.label:
yield s, p, Label(label=o, labelRoot=self.root).rdfs_label
yield s, skos.prefLabel, o
else:
yield s, p, o
class DKTLabels(LabelsBase):
""" Labels from the Desikan-Killiany-Tourville human brain labeling protocol. """
filename = 'dkt-labels'
name = 'Desikan-Killiany-Tourville Parcellation Labels'
shortname = 'dkt'
namespace = DKT
prefixes = {'DKTr':str(DKTr), 'DKTs':str(DKTs), **LabelsBase.prefixes}
imports = parcCore,
sources = DKTSrc,
root = LabelRoot(iri=nsExact(namespace), # ilxtr.hcpmmproot,
label='DKT label root',
shortname=shortname,
definingArtifacts=(s.artifact.iri for s in sources),
)
def _triples(self):
for source in self.sources:
for namespace, index, label in source:
iri = namespace[str(index)]
yield from Label(labelRoot=self.root,
label=label,
iri=iri)
yield from source.isVersionOf
class MNDBGLLabels():
""" DEPRECATED use FreeSurfer """
filename = 'mndbgl-labels'
name = 'Mindboggle Parcellation Labels'
shortname = 'mndbgl'
namespace = MNDBGL
imports = parcCore,
sources = MNDBGLSrc,
root = LabelRoot(iri=nsExact(namespace), # ilxtr.hcpmmproot,
label='Mindboggle label root',
shortname=shortname,
comment=('The local integer identifiers in the MNDBGL namespace '
'correspond to numbers from FreeSurferColorLUT.txt.'),
#definingArtifacts=(s.artifact.iri for s in sources),
)
def _triples(self):
for source in self.sources:
for index, label in source:
iri = self.namespace[str(index)]
yield from Label(labelRoot=self.root,
label=label,
iri=iri)
yield from source.isVersionOf
class PaxMouseLabels(PaxLabels):
""" Compilation of all labels used to name mouse brain regions
in atlases created using Paxinos and Franklin\'s methodology."""
# TODO FIXME align indexes where possible to paxrat???
filename = 'paxinos-mus-labels'
name = 'Paxinos & Franklin Mouse Parcellation Labels'
shortname = 'paxmus'
namespace = PAXMUS
prefixes = {
**makePrefixes('NIFRID', 'ilxtr', 'prov', 'dcterms'),
'PAXMUS': str(PAXMUS),
'paxmusver': str(paxmusver),
}
sources = PaxMFix, PaxMSrAr_2, PaxMSrAr_3
root = LabelRoot(
iri=nsExact(namespace), # PAXMUS['0'],
label='Paxinos mouse parcellation label root',
shortname=shortname,
definingArtifactsS=(Artifacts.PaxMouseAt.iri, ),
)
_merge = {
'4/5Cb': '4&5Cb',
'5N': 'Mo5',
'12N': '12',
'AngT': 'Ang',
'ANS': 'Acc',
'ASt': 'AStr',
'hif': 'hf',
'MnM': 'MMn',
'MoDG': 'Mol',
'och': 'ox',
'PHA': 'PH', # FIXME PH is reused in 3rd
'ST': 'BST',
'STIA': 'BSTIA',
'STLD': 'BSTLD',
'STLI': 'BSTLI',
'STLJ': 'BSTLJ',
'STLP': 'BSTLP',
'STLV': 'BSTLV',
'STMA': 'BSTMA',
'STMP': 'BSTMP',
'STMPI': 'BSTMPI',
'STMPL': 'BSTMPL',
'STMPM': 'BSTMPM',
'STMV': 'BSTMV',
'STS': 'BSTS',
}
class BermanLabels(LabelsBase):
""" Berman Cat labels """
# sort by label/structure not by abbrev
filename = 'berman-cat-labels'
name = 'Berman 1968 cat brain stem labels'
shortname = 'bercat'
imports = parcCore,
prefixes = {
**makePrefixes('NIFRID', 'ilxtr', 'prov'), 'BERCAT': str(BERCAT)
}
sources = BermanSrc,
namespace = BERCAT
root = LabelRoot(
iri=nsExact(namespace),
label='Berman 1968 cat label root',
shortname=shortname,
definingArtifacts=(s.artifact.iri for s in sources),
)
def _triples(self):
for source in self.sources:
for i, (label, paren_thing, abbrev, index) in enumerate(source):
local_identifier = str(i + 1)
iri = self.namespace[
local_identifier] # TODO load from existing
yield from Label(
labelRoot=self.root,
label=label,
#altLabel=None,
#synonyms=extras,
abbrevs=(abbrev, ),
iri=iri,
)
if paren_thing:
yield iri, ilx[
'berman/uris/readable/hasWeirdParenValue'], rdflib.Literal(
paren_thing)
continue
# FIXME different file ...
region_iri = ilx['berman/uris/cat/regions/' + local_identifier]
# FIXME incorporate version in tree or no?
# just have it be consecutive? HRM
yield region_iri, rdf.type, owl.Class
yield region_iri, ilxtr.hasParcellationLabel, iri # FIXME predicate choice ...
yield region_iri, ilxtr.isDefinedBy, BermanSrc.artifact.iri # FIXME
for plate_num in index:
yield region_iri, ilxtr.appearsOnPlateNumber, rdflib.Literal(
plate_num) # FIXME generalize ...
class WHSSDLabels(LabelsBase):
filename = 'waxholm-rat-labels'
name = 'Waxholm Sprague Dawley Atlas Labels'
shortname = 'whssd'
imports = parcCore,
prefixes = {
**makePrefixes('NIFRID', 'ilxtr', 'prov', 'dcterms'), 'WHSSD':
str(WHSSD)
}
sources = WHSSDSrc2, WHSSDilfSrc2, WHSSDSrc1, WHSSDilfSrc1
namespace = WHSSD
root = LabelRoot(
iri=nsExact(namespace), # ilxtr.whssdroot,
label='Waxholm Space Sprague Dawley parcellation label root',
shortname=shortname,
definingArtifacts=(s.artifact.iri for s in sources),
)
def _triples(self):
for source in self.sources:
preds = False
for index, label, *rest in source:
abbrev, parent = rest if rest else (
False, False) # a tricky one if you miss the parens
abbrevs = (abbrev, ) if abbrev and abbrev != label else tuple()
if int(index
) >= 1000: # FIXME this is the WRONG way to do this
# FIXME parentless structures in the ilf files?
label += ' (structure)'
iri = WHSSD[str(index)]
yield from Label(
labelRoot=self.root,
label=label,
abbrevs=abbrevs,
iri=iri,
)
if parent:
preds = True
parent = WHSSD[str(parent)]
yield from restriction.serialize(
iri, source.predicates[ilxtr.labelPartOf], parent)
yield from source.isVersionOf
if preds:
for parent, child in source.predicates.items(
): # FIXME annotationProperty vs objectProperty
yield child, rdfs.subPropertyOf, parent
class DMBALabels(HBALabels):
path = 'ttl/generated/parcellation/'
filename = 'dmbaslim'
name = 'Allen Developing Mouse Brain Atlas Ontology'
shortname = 'dmba'
prefixes = {
**makePrefixes('NIFRID', 'ilxtr', 'prov'), 'DMBA': str(DMBA),
'ilxDMBA': str(ilxDMBA)
}
sources = DMBASrc,
namespace = DMBA
root = LabelRoot(
iri=AIBS['mouse/devel/labels/'],
label='Allen Developing Mouse Brain Atlas parcellation label root',
shortname=shortname,
definingArtifacts=(s.artifact.iri for s in sources),
)
class FreeSurferLabels(LabelsBase):
filename = 'freesurfer-labels'
name = 'FreeSurfer color LUT Parcellation Labels'
shortname = 'fscl' # can't use fsl because that is already in use by another fMRI package...
namespace = FSCL
prefixes = {**LabelsBase.prefixes} # FIXME it would be nice to be able to automate this...
imports = parcCore,
sources = FreeSurferSrc,
root = LabelRoot(iri=nsExact(namespace), # ilxtr.hcpmmproot,
label='FreeSurfer Color LUT label root',
shortname=shortname,
definingArtifacts=(s.artifact.iri for s in sources),)
def _triples(self):
for source in self.sources:
for index, label in source:
iri = self.namespace[str(index)]
yield from Label(labelRoot=self.root,
label=label,
iri=iri)
yield from source.isVersionOf
class HBALabels(LabelsBase):
filename = 'hbaslim'
name = 'Allen Human Brain Atlas Ontology'
shortname = 'hba'
imports = parcCore,
prefixes = {
**makePrefixes('NIFRID', 'ilxtr', 'prov'), 'HBA': str(HBA),
'ilxHBA': str(ilxHBA)
}
sources = HBASrc,
namespace = HBA
root = LabelRoot(
iri=AIBS['human/labels/'], # ilxtr.hbaroot,
label='Allen Human Brain Atlas parcellation label root',
shortname=shortname,
definingArtifacts=(s.artifact.iri for s in sources),
)
def _triples(self):
for source in self.sources:
for record in source:
iri = self.namespace[str(record['id'])]
sn = record['safe_name']
if sn and sn != record['name']: syns = sn,
else: syns = tuple()
yield from Label(
labelRoot=self.root,
label=record['name'],
synonyms=syns,
abbrevs=(record['acronym'], ),
iri=iri,
)
superpart = record['parent_structure_id']
if superpart:
superpart_iri = self.namespace[str(superpart)]
yield from restriction.serialize(iri, ilxtr.labelPartOf,
superpart_iri)
def prepare(cls):
ATLAS_PATH = '/usr/share/fsl/data/atlases/'
shortnames = {
'JHU White-Matter Tractography Atlas': 'JHU WM',
'Oxford-Imanova Striatal Structural Atlas': 'OISS',
'Talairach Daemon Labels': 'Talairach',
'Subthalamic Nucleus Atlas': 'SNA',
'JHU ICBM-DTI-81 White-Matter Labels': 'JHU ICBM WM',
'Juelich Histological Atlas': 'Juelich',
'MNI Structural Atlas': 'MNI Struct',
}
prefixes = {
'Cerebellar Atlas in MNI152 space after normalization with FLIRT':
'CMNIfl',
'Cerebellar Atlas in MNI152 space after normalization with FNIRT':
'CMNIfn',
'Sallet Dorsal Frontal connectivity-based parcellation': 'DFCBP',
'Neubert Ventral Frontal connectivity-based parcellation': 'VFCBP',
'Mars Parietal connectivity-based parcellation': 'PCBP',
}
for xmlfile in glob.glob(ATLAS_PATH + '*.xml'):
filename = os.path.splitext(os.path.basename(xmlfile))[0]
tree = etree.parse(xmlfile)
parcellation_name = tree.xpath('header//name')[0].text
# namespace
namespace = rdflib.Namespace(FSLATS[filename + '/labels/'])
# shortname
shortname = tree.xpath('header//shortname')
if shortname:
shortname = shortname[0].text
else:
shortname = shortnames[parcellation_name]
artifact_shortname = shortname
shortname = shortname.replace(' ', '')
# Artifact
artifact = Terminology(
iri=FSLATS[filename],
label=parcellation_name,
docUri='http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/Atlases',
species=NCBITaxon['9606'],
devstage=UBERON[
'0000113'], # FIXME mature vs adult vs when they actually did it...
region=UBERON['0000955'],
shortname=artifact_shortname)
setattr(cls.Artifacts, shortname, artifact)
# LabelRoot
root = LabelRoot(iri=nsExact(namespace),
label=parcellation_name + ' label root',
shortname=shortname,
definingArtifacts=(artifact.iri, ))
root.namespace = namespace
cls.roots += root,
# prefix
if parcellation_name in prefixes:
prefix = 'fsl' + prefixes[parcellation_name]
else:
prefix = 'fsl' + shortname
cls.prefixes[prefix] = root.iri
# Source
@classmethod
def loadData(cls, _tree=tree):
out = []
for node in _tree.xpath('data//label'):
index, label = node.get('index'), node.text
out.append((index, label))
return out
source = type(
'FSLsource_' + shortname.replace(' ', '_'),
(Source, ),
dict(
iri=rdflib.URIRef('file://' + xmlfile),
source=xmlfile,
source_original=True,
artifact=artifact,
root=
root, # used locally since we have more than one root per ontology here
loadData=loadData))
cls.sources += source,
super().prepare()
class PaxRatLabels(PaxLabels):
""" Compilation of all labels used to name rat brain regions
in atlases created using Paxinos and Watson\'s methodology."""
filename = 'paxinos-rat-labels'
name = 'Paxinos & Watson Rat Parcellation Labels'
shortname = 'paxrat'
namespace = PAXRAT
prefixes = {
**makePrefixes('NIFRID', 'ilxtr', 'prov', 'dcterms'),
'PAXRAT': str(PAXRAT),
'paxratver': str(paxratver),
}
# sources need to go in the order with which we want the labels to take precedence (ie in this case 6e > 4e)
sources = PaxFix, PaxSrAr_6, PaxSr_6, PaxSrAr_4, PaxFix6, PaxFix4 #, PaxTree_6() # tree has been successfully used for crossreferencing, additional terms need to be left out at the moment (see in_tree_not_in_six)
root = LabelRoot(
iri=nsExact(namespace), # PAXRAT['0'],
label='Paxinos rat parcellation label root',
shortname=shortname,
#definingArtifactsS=None,#Artifacts.PaxRatAt.iri,
definingArtifactsS=(Artifacts.PaxRatAt.iri, ),
)
_fixes = []
_dupes = {
# for 4e the numbers in the index are to the cranial nerve nuclei entries
'3N':
DupeRecord(alt_abbrevs=['3'],
structures=['oculomotor nucleus'],
figures={},
artiris=[Artifacts.PaxRat4.iri]),
'4N':
DupeRecord(alt_abbrevs=['4'],
structures=['trochlear nucleus'],
figures={},
artiris=[Artifacts.PaxRat4.iri]),
'6N':
DupeRecord(alt_abbrevs=['6'],
structures=['abducens nucleus'],
figures={},
artiris=[Artifacts.PaxRat4.iri]),
'7N':
DupeRecord(alt_abbrevs=['7'],
structures=['facial nucleus'],
figures={},
artiris=[Artifacts.PaxRat4.iri]),
'10N':
DupeRecord(alt_abbrevs=['10'],
structures=['dorsal motor nucleus of vagus'],
figures={},
artiris=[Artifacts.PaxRat4.iri]),
# FIXME need comments about the index entries
'1Cb':
DupeRecord(alt_abbrevs=['1'],
structures=['cerebellar lobule 1'],
figures={},
artiris=[Artifacts.PaxRat4.iri]),
'2Cb':
DupeRecord(alt_abbrevs=['2'],
structures=['cerebellar lobule 2'],
figures={},
artiris=[Artifacts.PaxRat4.iri]),
'2/3Cb':
DupeRecord(alt_abbrevs=['2&3'],
structures=['cerebellar lobules 2&3'],
figures={},
artiris=[Artifacts.PaxRat4.iri]),
'3Cb':
DupeRecord(alt_abbrevs=['3'],
structures=['cerebellar lobule 3'],
figures={},
artiris=[Artifacts.PaxRat4.iri]),
'4Cb':
DupeRecord(alt_abbrevs=['4'],
structures=['cerebellar lobule 4'],
figures={},
artiris=[Artifacts.PaxRat4.iri]),
'4/5Cb':
DupeRecord(alt_abbrevs=['4&5'],
structures=['cerebellar lobules 4&5'],
figures={},
artiris=[Artifacts.PaxRat4.iri]),
'5Cb':
DupeRecord(alt_abbrevs=['5'],
structures=['cerebellar lobule 5'],
figures={},
artiris=[Artifacts.PaxRat4.iri]),
'6Cb':
DupeRecord(alt_abbrevs=['6'],
structures=['cerebellar lobule 6'],
figures={},
artiris=[Artifacts.PaxRat4.iri]),
'6aCb':
DupeRecord(alt_abbrevs=['6a'],
structures=['cerebellar lobule 6a'],
figures={},
artiris=[Artifacts.PaxRat4.iri]),
'6bCb':
DupeRecord(alt_abbrevs=['6b'],
structures=['cerebellar lobule 6b'],
figures={},
artiris=[Artifacts.PaxRat4.iri]),
'6cCb':
DupeRecord(alt_abbrevs=['6c'],
structures=['cerebellar lobule 6c'],
figures={},
artiris=[Artifacts.PaxRat4.iri]),
'7Cb':
DupeRecord(alt_abbrevs=['7'],
structures=['cerebellar lobule 7'],
figures={},
artiris=[Artifacts.PaxRat4.iri]),
'8Cb':
DupeRecord(alt_abbrevs=['8'],
structures=['cerebellar lobule 8'],
figures={},
artiris=[Artifacts.PaxRat4.iri]),
'9Cb':
DupeRecord(alt_abbrevs=['9'],
structures=['cerebellar lobule 9'],
figures={},
artiris=[Artifacts.PaxRat4.iri]),
'10Cb':
DupeRecord(alt_abbrevs=['10'],
structures=['cerebellar lobule 10'],
figures={},
artiris=[Artifacts.PaxRat4.iri]),
}
_merge = { # abbrevs that have identical structure names
'5N':'Mo5',
'12N':'12',
'ANS':'Acc',
'ASt':'AStr',
'AngT':'Ang',
'MnM':'MMn',
'MoDG':'Mol',
'PDPO':'PDP',
'PTg':'PPTg',
'STIA':'BSTIA',
'STL':'BSTL',
'STLD':'BSTLD',
'STLI':'BSTLI',
'STLJ':'BSTLJ',
'STLP':'BSTLP',
'STLV':'BSTLV',
'STM':'BSTM',
'STMA':'BSTMA',
'STMP':'BSTMP',
'STMPI':'BSTMPI',
'STMPL':'BSTMPL',
'STMPM':'BSTMPM',
'STMV':'BSTMV',
'hif':'hf',
'och':'ox',
}
def curate(self):
fr, err4 = PaxSrAr_4()
sx, err6 = PaxSrAr_6()
sx2, _ = PaxSr_6()
tr, err6t = PaxTree_6()
sfr = set(fr)
ssx = set(sx)
ssx2 = set(sx2)
str_ = set(tr)
in_four_not_in_six = sfr - ssx
in_six_not_in_four = ssx - sfr
in_tree_not_in_six = str_ - ssx
in_six_not_in_tree = ssx - str_
in_six2_not_in_six = ssx2 - ssx
in_six_not_in_six2 = ssx - ssx2
print(
len(in_four_not_in_six),
len(in_six_not_in_four),
len(in_tree_not_in_six),
len(in_six_not_in_tree),
len(in_six2_not_in_six),
len(in_six_not_in_six2),
)
tr_struct_abrv = {}
for abrv, ((struct, *extra), _, parent) in tr.items():
tr_struct_abrv[struct] = abrv
if abrv in sx:
#print(abrv, struct, parent)
if struct and struct not in sx[abrv][0]:
print(
f'Found new label from tr for {abrv}:\n{struct}\n{sx[abrv][0]}\n'
)
# can't run these for tr yet
#reduced = set(tr_struct_abrv.values())
#print(sorted(_ for _ in tr if _ not in reduced))
#assert len(tr_struct_abrv) == len(tr), 'mapping between abrvs and structs is not 1:1 for tr'
sx2_struct_abrv = {}
for abrv, ((struct, *extra), _) in sx2.items():
sx2_struct_abrv[struct] = abrv
if abrv in sx:
if struct and struct not in sx[abrv][0]:
print(
f'Found new label from sx2 for {abrv}:\n{struct}\n{sx[abrv][0]}\n'
)
reduced = set(sx2_struct_abrv.values())
print(sorted(
_ for _ in reduced
if _ not in sx2)) # ah inconsistent scoping rules in class defs...
assert len(sx2_struct_abrv) == len(sx2), 'there is a duplicate struct'
sx_struct_abrv = {}
for abrv, ((struct, *extra), _) in sx.items():
sx_struct_abrv[struct] = abrv
reduced = set(sx_struct_abrv.values())
print(sorted(_ for _ in reduced if _ not in sx))
assert len(sx_struct_abrv) == len(sx), 'there is a duplicate struct'
# TODO test whether any of the tree members that were are going to exclude have children that we are going to include
names_match_not_abbervs = {}
tree_no_name = {
_: tr[_]
for _ in sorted(in_tree_not_in_six) if not tr[_][0][0]
}
tree_with_name = {
_: tr[_]
for _ in sorted(in_tree_not_in_six) if tr[_][0][0]
}
not_in_tree_with_figures = {
_: sx[_]
for _ in sorted(in_six_not_in_tree) if sx[_][-1]
}
a = f'{"abv":<25} | {"structure name":<60} | parent abv\n' + '\n'.join(
f'{k:<25} | {v[0][0]:<60} | {v[-1]}'
for k, v in tree_with_name.items())
b = f'{"abv":<25} | {"structure name":<15} | parent abv\n' + '\n'.join(
f'{k:<25} | {"":<15} | {v[-1]}' for k, v in tree_no_name.items())
c = f'abv | {"structure name":<60} | figures (figure ranges are tuples)\n' + '\n'.join(
f'{k:<6} | {v[0][0]:<60} | {v[-1]}'
for k, v in not_in_tree_with_figures.items())
with open(
os.path.expanduser(
'~/ni/dev/nifstd/paxinos/tree-with-name.txt'), 'wt') as f:
f.write(a)
with open(
os.path.expanduser('~/ni/dev/nifstd/paxinos/tree-no-name.txt'),
'wt') as f:
f.write(b)
with open(
os.path.expanduser(
'~/ni/dev/nifstd/paxinos/not-in-tree-with-figures.txt'),
'wt') as f:
f.write(c)
#match_name_not_abrev = set(v[0][0] for v in tree_with_name.values()) & set(v[0][0] for v in sx.values())
_match_name_not_abrev = {}
for a, (alts, (s, *extra), f,
*_) in PaxRatLabels().records()[0].items():
if s not in _match_name_not_abrev:
_match_name_not_abrev[s] = [a]
elif a not in _match_name_not_abrev[s]:
_match_name_not_abrev[s].append(a)
match_name_not_abrev = {
k: v
for k, v in _match_name_not_abrev.items() if len(v) > 1
}
abrv_match_not_name = {
k: v[0]
for k, v in PaxRatLabels().records()[0].items() if len(v[0]) > 1
}
_ = [
print(k, *v[0]) for k, v in PaxRatLabels().records()[0].items()
if len(v[0]) > 1
]
breakpoint()