def __init__(self, genome, CoordName, Start, End, Strand = 1,
CoordType = None, seq_region_id = None, ensembl_coord=False):
if not CoordType or not (seq_region_id or Start or End):
seq_region_data, CoordType = \
_get_coord_type_and_seq_region_id(CoordName, genome.CoreDb)
seq_region_id = seq_region_data['seq_region_id']
Start = Start or 0
End = End or seq_region_data['length']
# TODO allow creation with just seq_region_id
self.Species = genome.Species
self.CoordType = DisplayString(CoordType, repr_length=4,
with_quotes=False)
self.CoordName = DisplayString(CoordName, repr_length=4,
with_quotes=False)
# if Start == End, we +1 to End, unless these are ensembl_coord's
if ensembl_coord:
Start -= 1
elif Start == End:
End += 1
if Start > End:
assert Strand == -1,\
"strand incorrect for start[%s] > end[%s]" % (Start, End)
Start, End = End, Start
self.Start = Start
self.End = End
self.Strand = convert_strand(Strand)
self.seq_region_id = seq_region_id
self.genome = genome
def getRegion(self,
region=None,
CoordName=None,
Start=None,
End=None,
Strand=None,
ensembl_coord=False):
"""returns a single generic region for the specified coordinates
Arguments:
- region: a genomic region or a Coordinate instance
- ensembl_coords: if True, follows indexing system of Ensembl
where indexing starts at 1"""
if region is None:
seq_region_id = self._get_seq_region_id(CoordName)
region = Coordinate(self,
CoordName=CoordName,
Start=Start,
End=End,
Strand=convert_strand(Strand),
seq_region_id=seq_region_id,
ensembl_coord=ensembl_coord)
elif hasattr(region, 'Location'):
region = region.Location
return GenericRegion(self,
self.CoreDb,
CoordName=CoordName,
Start=Start,
End=End,
Strand=Strand,
Location=region,
ensembl_coord=ensembl_coord)
def __init__(self, genome, CoordName, Start, End, Strand = 1,
CoordType = None, seq_region_id = None, ensembl_coord=False):
if not CoordType or not (seq_region_id or Start or End):
seq_region_data, CoordType = \
_get_coord_type_and_seq_region_id(CoordName, genome.CoreDb)
seq_region_id = seq_region_data['seq_region_id']
Start = Start or 0
End = End or seq_region_data['length']
# TODO allow creation with just seq_region_id
self.Species = genome.Species
self.CoordType = DisplayString(CoordType, repr_length=4,
with_quotes=False)
self.CoordName = DisplayString(CoordName, repr_length=4,
with_quotes=False)
# if Start == End, we +1 to End, unless these are ensembl_coord's
if ensembl_coord:
Start -= 1
elif Start == End:
End += 1
if Start > End:
assert Strand == -1,\
"strand incorrect for start[%s] > end[%s]" % (Start, End)
Start, End = End, Start
self.Start = Start
self.End = End
self.Strand = convert_strand(Strand)
self.seq_region_id = seq_region_id
self.genome = genome
def getRegion(self, region=None, CoordName=None, Start=None, End=None, Strand=None, ensembl_coord=False):
"""returns a single generic region for the specified coordinates
Arguments:
- region: a genomic region or a Coordinate instance
- ensembl_coords: if True, follows indexing system of Ensembl
where indexing starts at 1"""
if region is None:
seq_region_id = self._get_seq_region_id(CoordName)
region = Coordinate(
self,
CoordName=CoordName,
Start=Start,
End=End,
Strand=convert_strand(Strand),
seq_region_id=seq_region_id,
ensembl_coord=ensembl_coord,
)
elif hasattr(region, "Location"):
region = region.Location
return GenericRegion(
self,
self.CoreDb,
CoordName=CoordName,
Start=Start,
End=End,
Strand=Strand,
Location=region,
ensembl_coord=ensembl_coord,
)
def getFeatures(self,
region=None,
feature_types=None,
where_feature=None,
CoordName=None,
Start=None,
End=None,
Strand=None,
ensembl_coord=False):
"""returns Region instances for the specified location"""
if isinstance(feature_types, str):
feature_types = [feature_types]
feature_types = [ft.lower() for ft in feature_types]
feature_coord_levels = self._get_feature_coord_levels(feature_types)
if region is None:
seq_region_id = self._get_seq_region_id(CoordName)
region = Coordinate(self,
CoordName=CoordName,
Start=Start,
End=End,
Strand=convert_strand(Strand),
seq_region_id=seq_region_id,
ensembl_coord=ensembl_coord)
elif hasattr(region, 'Location'):
region = region.Location
coord = region
# the coordinate system at which locations are to be referenced, and
# the processing function
target_coords_funcs = \
dict(cpg = (self._get_simple_features, CpGisland),
repeat = (self._get_repeat_features, Repeat),
gene = (self._get_gene_features, Gene),
est = (self._get_gene_features, Est),
variation = (self._get_variation_features, Variation))
known_types = set(target_coords_funcs.keys())
if not set(feature_types) <= known_types:
raise RuntimeError, 'Unknown feature[%s], valid feature_types \
are: %s' % (set(feature_types) ^ known_types, known_types)
for feature_type in feature_types:
target_func, target_class = target_coords_funcs[feature_type]
db = self.CoreDb
if feature_type == 'est':
db = self.OtherFeaturesDb
feature_coords = feature_coord_levels[feature_type].levels
for feature_coord in feature_coords:
chrom_other_coords = get_coord_conversion(coord,
feature_coord,
db,
where=where_feature)
for chrom_coord, other_coord in chrom_other_coords:
for region in target_func(db, target_class, chrom_coord,
other_coord, where_feature):
yield region
def test_strand_conversion(self):
"""should consistently convert strand info"""
self.assertEquals(convert_strand(None), 1)
self.assertEquals(convert_strand(-1), -1)
self.assertEquals(convert_strand(1), 1)
self.assertEquals(convert_strand('-'), -1)
self.assertEquals(convert_strand('+'), 1)
self.assertEquals(convert_strand(-1.0), -1)
self.assertEquals(convert_strand(1.0), 1)
def test_strand_conversion(self):
"""should consistently convert strand info"""
self.assertEquals(convert_strand(None), 1)
self.assertEquals(convert_strand(-1), -1)
self.assertEquals(convert_strand(1), 1)
self.assertEquals(convert_strand('-'), -1)
self.assertEquals(convert_strand('+'), 1)
self.assertEquals(convert_strand(-1.0), -1)
self.assertEquals(convert_strand(1.0), 1)
def getFeatures(self, region=None, feature_types=None, where_feature=None,
CoordName=None, Start=None, End=None, Strand=None,
ensembl_coord=False):
"""returns Region instances for the specified location"""
if isinstance(feature_types, str):
feature_types = [feature_types]
feature_types = [ft.lower() for ft in feature_types]
feature_coord_levels = self._get_feature_coord_levels(feature_types)
if region is None:
seq_region_id = self._get_seq_region_id(CoordName)
region = Coordinate(self,CoordName=CoordName, Start=Start,
End=End,
Strand = convert_strand(Strand),
seq_region_id=seq_region_id,
ensembl_coord=ensembl_coord)
elif hasattr(region, 'Location'):
region = region.Location
coord = region
# the coordinate system at which locations are to be referenced, and
# the processing function
target_coords_funcs = \
dict(cpg = (self._get_simple_features, CpGisland),
repeat = (self._get_repeat_features, Repeat),
gene = (self._get_gene_features, Gene),
est = (self._get_gene_features, Est),
variation = (self._get_variation_features, Variation))
known_types = set(target_coords_funcs.keys())
if not set(feature_types) <= known_types:
raise RuntimeError, 'Unknown feature[%s], valid feature_types \
are: %s' % (set(feature_types)^known_types, known_types)
for feature_type in feature_types:
target_func, target_class = target_coords_funcs[feature_type]
db = self.CoreDb
if feature_type == 'est':
db = self.OtherFeaturesDb
feature_coords = feature_coord_levels[feature_type].levels
for feature_coord in feature_coords:
chrom_other_coords = get_coord_conversion(coord, feature_coord,
db, where=where_feature)
for chrom_coord, other_coord in chrom_other_coords:
for region in target_func(db, target_class, chrom_coord,
other_coord, where_feature):
yield region
