class Sample(db.Model):
id = db.Column(db.Integer, primary_key=True)
code = db.Column(db.String(32), index=True, unique=True)
family_code = db.Column(db.String(32), index=True)
gender = db.Column(db.String(32))
family_relationship = db.Column(db.String(128))
comments = db.Column(db.Text)
population_id = db.Column(db.Integer, db.ForeignKey('population.id'))
data_source_id = db.Column(db.Integer, db.ForeignKey('data_source.id'))
archaic_genome_data = db.relationship('ArchaicGenomeData',
backref='sample',
lazy='dynamic')
def __repr__(self):
return '<Sample {}>'.format(self.code)
class SuperPopulation(db.Model):
id = db.Column(db.Integer, primary_key=True)
code = db.Column(db.String(32), index=True, unique=True)
name = db.Column(db.String(128))
color = db.Column(db.String(32))
data_source_id = db.Column(db.Integer, db.ForeignKey('data_source.id'))
populations = db.relationship('Population',
secondary=population_group_table,
backref=db.backref('super_populations',
lazy=True))
# TODO Make this a hybrid_property?
def samples(self):
samples = set()
for population in self.populations:
for sample in population.samples:
samples.add(sample)
return samples
@property
def population_count(self):
return db.Session.object_session(self).\
query(Population).with_parent(self, "populations").count()
@property
def sample_count(self):
return db.Session.object_session(self).\
query(Sample).join(Population).\
with_parent(self, "populations").count()
def __repr__(self):
return '<SuperPopulation {}>'.format(self.code)
class ArchaicAnalysisRun(db.Model):
id = db.Column(db.Integer, primary_key=True)
name = db.Column(db.String(128), index=True, unique=True)
description = db.Column(db.Text)
publication_doi = db.Column(db.String(256))
date = db.Column(db.DateTime)
data_source_id = db.Column(db.Integer, db.ForeignKey('data_source.id'))
archaic_genome_data = db.relationship('ArchaicGenomeData',
backref='archaic_analysis_run',
lazy='dynamic')
@hybrid_property
def publication_url(self):
publication_url = None
if self.publication_doi is not None:
publication_url = "https://doi.org/{}".foramt(self.publication_doi)
return publication_url
@hybrid_property
def sampleids_with_data_query(self):
return db.session.query(Sample.id).join(ArchaicGenomeData).\
join(ArchaicAnalysisRun).\
filter(ArchaicAnalysisRun.id == self.id)
@hybrid_property
def samples_with_data_query(self):
return Sample.query.filter(
Sample.id.in_(self.sampleids_with_data_query))
@hybrid_property
def samples_with_data_count(self):
return self.samples_with_data_query.count()
@hybrid_property
def samples_with_data(self):
return self.samples_with_data_query.all()
@hybrid_property
def samples_without_data_query(self):
return Sample.query.filter(
Sample.id.notin_(self.sampleids_with_data_query))
@hybrid_property
def samples_without_data_count(self):
return self.samples_without_data_query.count()
@hybrid_property
def samples_without_data(self):
return self.samples_without_data_query.all()
def __repr__(self):
return '<ArchaicAnalysisRun {}>'.format(self.name)
class ArchaicGenomeData(db.Model):
id = db.Column(db.Integer, primary_key=True)
sample_id = db.Column(db.Integer,
db.ForeignKey('sample.id'),
nullable=False)
archaic_analysis_run_id = db.Column(
db.Integer, db.ForeignKey('archaic_analysis_run.id'), nullable=False)
archaic_genome_call = db.Column(db.String(128), nullable=False, index=True)
haplotype = db.Column(db.Integer, nullable=False)
__table_args__ = (db.Index('idx_sample_run_genome_haplotype',
'sample_id',
'archaic_analysis_run_id',
'archaic_genome_call',
'haplotype',
unique=True), )
bed_file = db.Column(db.String(512), nullable=False)
total_bps = db.Column(db.Integer, nullable=False)
total_haplotypes = db.Column(db.Integer, nullable=False)
def __repr__(self):
return '<ArchaicGenomeData {}:{}:{}:{}>'.format(
self.sample.code, self.archaic_analysis_run.name,
self.archaic_genome_call, self.haplotype)
class Population(db.Model):
id = db.Column(db.Integer, primary_key=True)
code = db.Column(db.String(32), index=True, unique=True)
name = db.Column(db.String(128), unique=True)
description = db.Column(db.String(256))
latitude = db.Column(db.Float)
longitude = db.Column(db.Float)
data_source_id = db.Column(db.Integer, db.ForeignKey('data_source.id'))
samples = db.relationship('Sample', backref='population', lazy='dynamic')
sample_count = column_property(
select([func.count(Sample.id)
]).where(Sample.population_id == id).correlate_except(Sample))
@property
def archaic_analysis_runs(self):
return ArchaicAnalysisRun.query.\
join(ArchaicGenomeData).join(Sample).join(Population).\
filter(Population.id == self.id).all()
def archaic_genome_data_for_run_query(self, archaic_analysis_run_id):
return ArchaicGenomeData.query.join(ArchaicAnalysisRun).\
join(Sample).join(Population).filter(Population.id == self.id).\
filter(ArchaicAnalysisRun.id == archaic_analysis_run_id)
def archaic_genome_data_for_run(self, archaic_analysis_run_id):
return self.archaic_genome_data_for_run_query(
archaic_analysis_run_id=archaic_analysis_run_id).all()
def samples_with_data_for_run_query(self, archaic_analysis_run_id):
return Sample.query.join(ArchaicGenomeData).\
join(ArchaicAnalysisRun).\
join(Population).\
filter(ArchaicAnalysisRun.id == archaic_analysis_run_id).\
filter(Population.id == self.id)
def samples_with_data_for_run(self, archaic_analysis_run_id):
return self.samples_with_data_for_run_query(
archaic_analysis_run_id=archaic_analysis_run_id).all()
def avg_archaic_genome_stats(self, archaic_analysis_run_id):
stmt = self.archaic_genome_data_for_run_query(
archaic_analysis_run_id).subquery()
return archaic_genome_stats_avg(stmt)
def __repr__(self):
return '<Population {}>'.format(self.code)
class ArchaicAnalysisRun(db.Model):
id = db.Column(db.Integer, primary_key=True)
name = db.Column(db.String(128), index=True, unique=True)
description = db.Column(db.Text)
publication_doi = db.Column(db.String(256))
date = db.Column(db.DateTime)
data_source_id = db.Column(db.Integer, db.ForeignKey('data_source.id'))
archaic_genome_data = db.relationship('ArchaicGenomeData',
backref='archaic_analysis_run',
lazy='dynamic')
@hybrid_property
def publication_url(self):
publication_url = None
if self.publication_doi is not None:
publication_url = "https://doi.org/{}".foramt(self.publication_doi)
return publication_url
@hybrid_property
def sampleids_with_data_query(self):
return db.session.query(Sample.id).join(ArchaicGenomeData).\
join(ArchaicAnalysisRun).\
filter(ArchaicAnalysisRun.id == self.id)
@hybrid_property
def samples_with_data_query(self):
return Sample.query.filter(
Sample.id.in_(self.sampleids_with_data_query))
@hybrid_property
def samples_with_data_count(self):
return self.samples_with_data_query.count()
@hybrid_property
def samples_with_data(self):
return self.samples_with_data_query.all()
@hybrid_property
def samples_without_data_query(self):
return Sample.query.filter(
Sample.id.notin_(self.sampleids_with_data_query))
@hybrid_property
def samples_without_data_count(self):
return self.samples_without_data_query.count()
@hybrid_property
def samples_without_data(self):
return self.samples_without_data_query.all()
def get_statistics(self, population_id):
sums = db.session.query(ArchaicGenomeData.archaic_genome_call,
ArchaicGenomeData.haplotype,
func.sum(ArchaicGenomeData.total_bps).
label('total_bp'),
func.sum(ArchaicGenomeData.total_haplotypes).
label('total_haplotypes')).\
join(Sample).\
filter(Sample.population_id == population_id,
ArchaicGenomeData.archaic_analysis_run_id == self.id).\
group_by(ArchaicGenomeData.archaic_genome_call,
ArchaicGenomeData.haplotype).all()
result = {
'neandertal_bp': 0,
'neandertal_haplotypes': 0,
'denisovan_bp': 0,
'denisovan_haplotypes': 0,
}
total_bp = sum([s[2] for s in sums])
total_hap = sum([s[3] for s in sums])
for s in sums:
if s[0] == 'den':
result['denisovan_bp'] += s[2] / total_bp
result['denisovan_haplotypes'] += s[3] / total_hap
elif s[0] == 'neand':
result['neandertal_bp'] += s[2] / total_bp
result['neandertal_haplotypes'] += s[3] / total_hap
return result
def __repr__(self):
return '<ArchaicAnalysisRun {}>'.format(self.name)
family_relationship = db.Column(db.String(128))
comments = db.Column(db.Text)
population_id = db.Column(db.Integer, db.ForeignKey('population.id'))
data_source_id = db.Column(db.Integer, db.ForeignKey('data_source.id'))
archaic_genome_data = db.relationship('ArchaicGenomeData',
backref='sample',
lazy='dynamic')
def __repr__(self):
return '<Sample {}>'.format(self.code)
population_group_table = db.Table(
'population_group', db.Model.metadata,
db.Column('super_population_id', db.Integer,
db.ForeignKey('super_population.id')),
db.Column('population_id', db.Integer, db.ForeignKey('population.id')))
class Population(db.Model):
id = db.Column(db.Integer, primary_key=True)
code = db.Column(db.String(32), index=True, unique=True)
name = db.Column(db.String(128), unique=True)
description = db.Column(db.String(256))
latitude = db.Column(db.Float)
longitude = db.Column(db.Float)
data_source_id = db.Column(db.Integer, db.ForeignKey('data_source.id'))
samples = db.relationship('Sample', backref='population', lazy='dynamic')
sample_count = column_property(
select([func.count(Sample.id)