def format_attribute(self):
"""
Format an object instance as a dictionary used by biosample, for
example::
[{
'value': 'submitter',
'terms': [{'url': 'http://www.ebi.ac.uk/efo/EFO_0001741'}]
}]
the fixed part of URI link is defined by ``library_name`` class
attribute
"""
if self.library_name is None:
logger.warning("library_name not defined")
library_uri = OBO_URL
else:
library = Ontology.objects.get(library_name=self.library_name)
library_uri = library.library_uri
return format_attribute(value=self.label,
library_uri=library_uri,
terms=self.term)
def format_attribute(self):
"""Format mapped_breed attribute (with its ontology). Return None if
no mapped_breed"""
if not self.mapped_breed or not self.mapped_breed_term:
return None
library = Ontology.objects.get(library_name=self.library_name)
library_uri = library.library_uri
return format_attribute(value=self.mapped_breed,
library_uri=library_uri,
terms=self.mapped_breed_term)
def get_attributes(self):
"""Return attributes like biosample needs"""
attributes = super().get_attributes()
attributes["Material"] = format_attribute(value="organism",
terms="OBI_0100026")
# TODO: how to model derived from (mother/father)?
attributes['Supplied breed'] = format_attribute(
value=self.breed.supplied_breed)
# HINT: Ideally, I could retrieve an ontology id for countries
attributes['EFABIS Breed country'] = format_attribute(
value=self.breed.country.label)
attributes['Mapped breed'] = self.breed.format_attribute()
attributes['Sex'] = self.sex.format_attribute()
# a datetime object should be not be converted in string here,
# otherwise will not be filtered if NULL
attributes['Birth date'] = format_attribute(value=self.birth_date,
units="YYYY-MM-DD")
attributes["Birth location"] = format_attribute(
value=self.birth_location)
attributes["Birth location longitude"] = format_attribute(
value=self.birth_location_longitude, units="decimal degrees")
attributes["Birth location latitude"] = format_attribute(
value=self.birth_location_latitude, units="decimal degrees")
attributes["Birth location accuracy"] = format_attribute(
value=self.get_birth_location_accuracy_display())
# filter out empty values
attributes = {k: v for k, v in attributes.items() if v is not None}
return attributes
def get_attributes(self):
"""Common attribute definition required from Animal and samples. Need
to be called inside Animal/sample get_atribute method. Keys
is the name in metadata rules
Returns:
dict: a dictionary object
"""
attributes = {}
attributes['Data source ID'] = format_attribute(
value=self.data_source_id)
attributes['Alternative id'] = format_attribute(
value=self.alternative_id)
# HINT: this is a mandatory biosample field: could be removed from
# attributes?
attributes['Description'] = format_attribute(value=self.description)
attributes["Project"] = format_attribute(value="IMAGE")
# to retrieve where this sample belongs
attributes["IMAGE submission id"] = format_attribute(
value=self.submission.id)
attributes['Submission title'] = format_attribute(
value=self.submission.title)
attributes['Submission description'] = format_attribute(
value=self.submission.description)
attributes['Person last name'] = format_attribute(
value=self.owner.last_name)
attributes['Person initial'] = format_attribute(
value=self.person.initials)
attributes['Person first name'] = format_attribute(
value=self.owner.first_name)
attributes['Person email'] = format_attribute(value="mailto:%s" %
(self.owner.email))
attributes['Person affiliation'] = format_attribute(
value=self.person.affiliation.name)
attributes['Person role'] = self.person.role.format_attribute()
attributes['Organization name'] = format_attribute(
value=self.organization.name)
attributes['Organization address'] = format_attribute(
value=self.organization.address)
attributes['Organization uri'] = format_attribute(
value=self.organization.URI)
attributes['Organization country'] = \
self.organization.country.format_attribute()
attributes[
'Organization role'] = self.organization.role.format_attribute()
# this could be present or not
if self.publication:
attributes['Publication DOI'] = format_attribute(
value=self.publication.doi)
attributes['Gene bank name'] = format_attribute(
value=self.gene_bank_name)
attributes[
'Gene bank country'] = self.gene_bank_country.format_attribute()
attributes['Data source type'] = format_attribute(
value=self.submission.get_datasource_type_display())
attributes['Data source version'] = format_attribute(
value=self.submission.datasource_version)
attributes['Species'] = self.specie.format_attribute()
return attributes
def get_orphan_samples(limit=None):
"""
Iterate for all BioSample orphaned records which are not yet removed and
are tracked for removal, get minimal data from BioSample and return a
dictionary which can be used to patch a BioSample id with a new
BioSample submission in order to remove a BioSamples record
(publish the BioSample record after 1000 years from Now).
Yields
------
new_data : dict
payload to submit to BioSample in order to remove a BioSamples record.
"""
with requests.Session() as session:
# get all biosamples candidate for a removal. Pay attention that
# could be removed from different users
qs = ORPHAN_QS.order_by('team__name', 'id')
if limit:
qs = islice(qs, limit)
for orphan_sample in qs:
# define the url I need to check
url = "/".join([BIOSAMPLE_URL, orphan_sample.biosample_id])
# read data from url
response = session.get(url)
data = response.json()
# check status
if response.status_code == 403:
logger.error("Error for %s (%s): %s" % (
orphan_sample.biosample_id,
data['error'],
data['message'])
)
# this sample seems already removed
continue
# I need a new data dictionary to submit
new_data = dict()
# I suppose the accession exists, since I found this sample
# using accession [biosample.id]
new_data['accession'] = data.get(
'accession', orphan_sample.biosample_id)
new_data['alias'] = data['name']
new_data['title'] = data['characteristics']['title'][0]['text']
# this will be the most important attribute
new_data['releaseDate'] = str(
parse_date(data['releaseDate']) + RELEASE_TIMEDELTA)
new_data['taxonId'] = data['taxId']
# need to determine taxon as
new_data['taxon'] = DictSpecie.objects.get(
term__endswith=data['taxId']).label
new_data['attributes'] = dict()
new_data['description'] = "Removed by InjectTool"
# set project again
new_data['attributes']["Project"] = format_attribute(
value="IMAGE")
# return new biosample data
yield {
'data': new_data,
'team': orphan_sample.team,
'sample': orphan_sample,
}
def get_attributes(self):
"""Return attributes like biosample needs"""
attributes = super().get_attributes()
attributes["Material"] = format_attribute(
value="specimen from organism", terms="OBI_0001479")
# The data source id or alternative id of the animal from which
# the sample was collected (see Animal.to_biosample())
attributes['Derived from'] = format_attribute(value=self.animal.name)
attributes["Specimen collection protocol"] = format_attribute(
value=self.protocol)
# a datetime object should be not be converted in string here,
# otherwise will not be filtered if NULL
attributes['Collection date'] = format_attribute(
value=self.collection_date, units="YYYY-MM-DD")
attributes['Collection place'] = format_attribute(
value=self.collection_place)
attributes["Collection place longitude"] = format_attribute(
value=self.collection_place_longitude, units="decimal degrees")
attributes["Collection place latitude"] = format_attribute(
value=self.collection_place_latitude, units="decimal degrees")
attributes["Collection place accuracy"] = format_attribute(
value=self.get_collection_place_accuracy_display())
# this will point to a correct term dictionary table
if self.organism_part:
attributes['Organism part'] = self.organism_part.format_attribute()
if self.developmental_stage:
attributes['Developmental stage'] = \
self.developmental_stage.format_attribute()
if self.physiological_stage:
attributes['Physiological stage'] = \
self.physiological_stage.format_attribute()
attributes['Animal age at collection'] = format_attribute(
value=self.animal_age_at_collection,
units=self.get_animal_age_at_collection_units_display())
attributes['Availability'] = format_attribute(value=self.availability)
attributes['Sample storage'] = format_attribute(
value=self.get_storage_display())
attributes['Sample storage processing'] = format_attribute(
value=self.get_storage_processing_display())
attributes['Sampling to preparation interval'] = format_attribute(
value=self.preparation_interval,
units=self.get_preparation_interval_units_display())
# filter out empty values
attributes = {k: v for k, v in attributes.items() if v is not None}
return attributes