def as_orange_table_v3(self):
import numpy
import Orange.data
data = self.run(count=False, header=True)
data = data.decode("utf-8")
if self.format.lower() == "tsv":
header, data = data.split("\n", 1)
domain = Orange.data.Domain([], [], [
Orange.data.StringVariable(name) for name in header.split("\t")
])
rows = [
line.split("\t") for line in data.split("\n") if line.strip()
]
rows = numpy.array(rows, dtype=object)
X = numpy.empty((len(rows), 0))
return Orange.data.Table.from_numpy(domain, X, metas=rows)
elif self.format.lower() == "fasta":
from Bio import SeqIO
domain = Orange.data.Domain([], [], [
Orange.data.StringVariable("id"),
Orange.data.StringVariable("sequence")
])
rows = [[seq.id, str(seq.seq)]
for seq in SeqIO.parse(io.StringIO(data), "fasta")]
rows = numpy.array(rows, dtype=object)
X = numpy.empty((len(rows), 0))
return Orange.data.Table.from_numpy(domain, X, metas=rows)
else:
raise BioMartError("Unsupported format: %s" % self.format)
def as_orange_table_v3(self):
import numpy
import Orange.data
from Bio import SeqIO
data = self.run(count=False, header=True)
data = data.decode("utf-8")
if self.format.lower() == "tsv":
header, data = data.split("\n", 1)
domain = Orange.data.Domain(
[], [], [Orange.data.StringVariable(name)
for name in header.split("\t")])
rows = [line.split("\t")
for line in data.split("\n") if line.strip()]
rows = numpy.array(rows, dtype=object)
X = numpy.empty((len(rows), 0))
return Orange.data.Table.from_numpy(domain, X, metas=rows)
elif self.format.lower() == "fasta":
domain = Orange.data.Domain(
[], [],
[Orange.data.StringVariable("id"),
Orange.data.StringVariable("sequence")])
rows = [[seq.id, str(seq.seq)]
for seq in SeqIO.parse(io.StringIO(data), "fasta")]
rows = numpy.array(rows, dtype=object)
X = numpy.empty((len(rows), 0))
return Orange.data.Table.from_numpy(domain, X, metas=rows)
else:
raise BioMartError("Unsupported format: %s" % self.format)
def get_example_table(self):
import Orange.data
import Orange.feature
data = self.run(count=False, header=True)
if self.format.lower() == "tsv":
header, data = data.split("\n", 1)
domain = Orange.data.Domain(
[Orange.feature.String(name) for name in header.split("\t")],
None)
data = [
line.split("\t") for line in data.split("\n") if line.strip()
]
return Orange.data.Table(domain, data) if data else None
elif self.format.lower() == "fasta":
from Bio import SeqIO
domain = Orange.data.Domain([
Orange.feature.String("id"),
Orange.feature.String("sequence")
], False) # TODO: meaningful id
examples = []
for seq in SeqIO.parse(io.BytesIO(data), "fasta"):
examples.append([seq.id, str(seq.seq)])
return Orange.data.Table(domain, examples)
else:
raise BioMartError("Unsupported format: %s" % self.format)
def get_example_table(self):
import Orange.data
import Orange.feature
from Bio import SeqIO
data = self.run(count=False, header=True)
if self.format.lower() == "tsv":
header, data = data.split("\n", 1)
domain = Orange.data.Domain(
[Orange.feature.String(name) for name in header.split("\t")],
None)
data = [line.split("\t")
for line in data.split("\n") if line.strip()]
return Orange.data.Table(domain, data) if data else None
elif self.format.lower() == "fasta":
domain = Orange.data.Domain(
[Orange.feature.String("id"),
Orange.feature.String("sequence")],
False) # TODO: meaningful id
examples = []
for seq in SeqIO.parse(io.BytesIO(data), "fasta"):
examples.append([seq.id, str(seq.seq)])
return Orange.data.Table(domain, examples)
else:
raise BioMartError("Unsupported format: %s" % self.format)
