def _get_gene_features(self, db, klass, target_coord, query_coord,
where_feature):
"""returns all genes"""
xref_table = [None, db.getTable('xref')][db.Type == 'core']
gene_table = db.getTable('gene')
# after release 65, the gene_id_table is removed. The following is to maintain
# support for earlier releases.
if self.GeneralRelease >= 65:
gene_id_table = None
else:
gene_id_table = db.getTable('gene_stable_id')
# note gene records are at chromosome, not contig, level
condition = gene_table.c.seq_region_id == query_coord.seq_region_id
query = self._build_gene_query(db, condition, gene_table,
gene_id_table, xref_table)
query = location_query(gene_table,
query_coord.EnsemblStart,
query_coord.EnsemblEnd,
query=query,
where=where_feature)
for record in query.execute():
new = Coordinate(self,
CoordName=query_coord.CoordName,
Start=record['seq_region_start'],
End=record['seq_region_end'],
Strand=record['seq_region_strand'],
seq_region_id=record['seq_region_id'],
ensembl_coord=True)
gene = klass(self, db, Location=new, data=record)
yield gene
def _get_gene_features(self, db, klass, target_coord, query_coord,
where_feature):
"""returns all genes"""
xref_table = [None, db.getTable('xref')][db.Type == 'core']
gene_table = db.getTable('gene')
# after release 65, the gene_id_table is removed. The following is to maintain
# support for earlier releases.
if self.GeneralRelease >= 65:
gene_id_table = None
else:
gene_id_table = db.getTable('gene_stable_id')
# note gene records are at chromosome, not contig, level
condition = gene_table.c.seq_region_id == query_coord.seq_region_id
query = self._build_gene_query(db, condition, gene_table, gene_id_table, xref_table)
query = location_query(gene_table, query_coord.EnsemblStart,
query_coord.EnsemblEnd, query=query, where=where_feature)
for record in query.execute():
new = Coordinate(self, CoordName=query_coord.CoordName,
Start=record['seq_region_start'],
End = record['seq_region_end'],
Strand = record['seq_region_strand'],
seq_region_id=record['seq_region_id'],
ensembl_coord=True)
gene = klass(self, db, Location=new, data=record)
yield gene
def _get_repeat_features(self, db, klass, target_coord, query_coord,
where_feature):
"""returns Repeat region instances"""
# we build repeats using coordinates from repeat_feature table
# the repeat_consensus_id is required to get the repeat name, class
# and type
repeat_feature_table = db.getTable('repeat_feature')
query = sql.select(
[repeat_feature_table],
repeat_feature_table.c.seq_region_id == query_coord.seq_region_id)
query = location_query(repeat_feature_table,
query_coord.EnsemblStart,
query_coord.EnsemblEnd,
query=query,
where=where_feature)
for record in query.execute():
coord = Coordinate(self,
CoordName=query_coord.CoordName,
Start=record['seq_region_start'],
End=record['seq_region_end'],
seq_region_id=record['seq_region_id'],
Strand=record['seq_region_strand'],
ensembl_coord=True)
if query_coord.CoordName != target_coord.CoordName:
coord = asserted_one(
get_coord_conversion(coord, target_coord.CoordType,
self.CoreDb))[1]
# coord = coord.makeRelativeTo(query_coord) #TODO: fix here if query_coord and target_coord have different coordName
# coord = coord.makeRelativeTo(target_coord, False)
yield klass(self,
db,
Location=coord,
Score=record['score'],
data=record)
def _get_simple_features(self, db, klass, target_coord, query_coord,
where_feature):
"""returns feature_type records for the query_coord from the
simple_feature table. The returned coord is referenced to
target_coord. At present, only CpG islands being queried."""
simple_feature_table = db.getTable('simple_feature')
feature_types = ['CpGisland']
feature_type_ids=[str(self._feature_type_ids.get(f)) for f in feature_types]
# fix the following
query = sql.select([simple_feature_table],
sql.and_(simple_feature_table.c.analysis_id.in_(feature_type_ids),
simple_feature_table.c.seq_region_id == query_coord.seq_region_id))
query = location_query(simple_feature_table,query_coord.EnsemblStart,
query_coord.EnsemblEnd, query=query,
where=where_feature)
records = query.execute()
for record in records:
coord = Coordinate(self, CoordName=query_coord.CoordName,
Start=record['seq_region_start'],
End = record['seq_region_end'],
seq_region_id=record['seq_region_id'],
Strand = record['seq_region_strand'],
ensembl_coord=True)
if query_coord.CoordName != target_coord.CoordName:
coord = asserted_one(get_coord_conversion(coord, target_coord.CoordType, self.CoreDb))[1]
# coord = coord.makeRelativeTo(query_coord) #TODO: fix here if query_coord and target_coord have different coordName
# coord = coord.makeRelativeTo(target_coord, False)
yield klass(self, db, Location=coord, Score=record['score'])
def _get_repeat_features(self, db, klass, target_coord, query_coord, where_feature):
"""returns Repeat region instances"""
# we build repeats using coordinates from repeat_feature table
# the repeat_consensus_id is required to get the repeat name, class
# and type
repeat_feature_table = db.getTable("repeat_feature")
query = sql.select([repeat_feature_table], repeat_feature_table.c.seq_region_id == query_coord.seq_region_id)
query = location_query(
repeat_feature_table, query_coord.EnsemblStart, query_coord.EnsemblEnd, query=query, where=where_feature
)
for record in query.execute():
coord = Coordinate(
self,
CoordName=query_coord.CoordName,
Start=record["seq_region_start"],
End=record["seq_region_end"],
seq_region_id=record["seq_region_id"],
Strand=record["seq_region_strand"],
ensembl_coord=True,
)
if query_coord.CoordName != target_coord.CoordName:
coord = asserted_one(get_coord_conversion(coord, target_coord.CoordType, self.CoreDb))[1]
# coord = coord.makeRelativeTo(query_coord) #TODO: fix here if query_coord and target_coord have different coordName
# coord = coord.makeRelativeTo(target_coord, False)
yield klass(self, db, Location=coord, Score=record["score"], data=record)
def _get_variation_features(self, db, klass, target_coord, query_coord, where_feature):
"""returns variation instances within the specified region"""
# variation features at supercontig level
var_feature_table = self.VarDb.getTable("variation_feature")
# note gene records are at chromosome, not contig, level
query = sql.select([var_feature_table], var_feature_table.c.seq_region_id == query_coord.seq_region_id)
query = location_query(
var_feature_table, query_coord.EnsemblStart, query_coord.EnsemblEnd, query=query, where=where_feature
)
for record in query.execute():
yield klass(self, self.CoreDb, Symbol=record["variation_name"], data=record)
def _get_genomic_align_blocks_for_dna_frag_id(self, method_clade_id,
dnafrag_id, coord):
genomic_align_table = self.ComparaDb.getTable('genomic_align')
query = sql.select([genomic_align_table.c.genomic_align_id,
genomic_align_table.c.genomic_align_block_id],
sql.and_(genomic_align_table.c.method_link_species_set_id ==\
method_clade_id,
genomic_align_table.c.dnafrag_id == dnafrag_id))
query = location_query(genomic_align_table,
coord.EnsemblStart,
coord.EnsemblEnd,
start_col='dnafrag_start',
end_col='dnafrag_end',
query=query)
return query.execute().fetchall()
def _get_genomic_align_blocks_for_dna_frag_id(self, method_clade_id,
dnafrag_id, coord):
genomic_align_table = self.ComparaDb.getTable('genomic_align')
query = sql.select([genomic_align_table.c.genomic_align_id,
genomic_align_table.c.genomic_align_block_id],
sql.and_(genomic_align_table.c.method_link_species_set_id ==\
method_clade_id,
genomic_align_table.c.dnafrag_id == dnafrag_id))
query = location_query(genomic_align_table,
coord.EnsemblStart,
coord.EnsemblEnd,
start_col = 'dnafrag_start',
end_col = 'dnafrag_end',
query = query)
return query.execute().fetchall()
def _get_variation_features(self, db, klass, target_coord, query_coord,
where_feature):
"""returns variation instances within the specified region"""
# variation features at supercontig level
var_feature_table = self.VarDb.getTable('variation_feature')
# note gene records are at chromosome, not contig, level
query = sql.select(
[var_feature_table],
var_feature_table.c.seq_region_id == query_coord.seq_region_id)
query = location_query(var_feature_table,
query_coord.EnsemblStart,
query_coord.EnsemblEnd,
query=query,
where=where_feature)
for record in query.execute():
yield klass(self,
self.CoreDb,
Symbol=record['variation_name'],
data=record)
def _get_simple_features(self, db, klass, target_coord, query_coord,
where_feature):
"""returns feature_type records for the query_coord from the
simple_feature table. The returned coord is referenced to
target_coord. At present, only CpG islands being queried."""
simple_feature_table = db.getTable('simple_feature')
feature_types = ['CpGisland']
feature_type_ids = [
self._feature_type_ids.get(f) for f in feature_types
]
# fix the following
query = sql.select(
[simple_feature_table],
sql.and_(
simple_feature_table.c.analysis_id.in_(feature_type_ids),
simple_feature_table.c.seq_region_id ==
query_coord.seq_region_id))
query = location_query(simple_feature_table,
query_coord.EnsemblStart,
query_coord.EnsemblEnd,
query=query,
where=where_feature)
records = query.execute()
for record in records:
coord = Coordinate(self,
CoordName=query_coord.CoordName,
Start=record['seq_region_start'],
End=record['seq_region_end'],
seq_region_id=record['seq_region_id'],
Strand=record['seq_region_strand'],
ensembl_coord=True)
if query_coord.CoordName != target_coord.CoordName:
coord = asserted_one(
get_coord_conversion(coord, target_coord.CoordType,
self.CoreDb))[1]
# coord = coord.makeRelativeTo(query_coord) #TODO: fix here if query_coord and target_coord have different coordName
# coord = coord.makeRelativeTo(target_coord, False)
yield klass(self, db, Location=coord, Score=record['score'])
def _get_gene_features(self, db, klass, target_coord, query_coord,
where_feature):
"""returns all genes"""
xref_table = [None, db.getTable('xref')][db.Type == 'core']
gene_table = db.getTable('gene')
# note gene records are at chromosome, not contig, level
condition = gene_table.c.seq_region_id == query_coord.seq_region_id
query = self._build_gene_query(db, condition, gene_table, xref_table)
query = location_query(gene_table, query_coord.EnsemblStart,
query_coord.EnsemblEnd, query=query, where=where_feature)
for record in query.execute():
new = Coordinate(self, CoordName=query_coord.CoordName,
Start=record['seq_region_start'],
End = record['nd'],
Strand = record['seq_region_strand'],
seq_region_id=record['seq_region_id'],
ensembl_coord=True)
if query_coord.CoordName != target_coord.CoordName:
coord = asserted_one(get_coord_conversion(coord, target_coord.CoordType, self.CoreDb))[1]
# TODO: check coord, used 'new' here. where is coord (above line) used?
gene = klass(self, db, Location=new, data=record)
yield gene
