def build(self, database_connection, **kwargs):
kwargs.setdefault('hypothesis_name', self.hypothesis_metadata['name'])
task = TextFileGlycanHypothesisSerializer(
self.prepare_buffer(), database_connection,
**kwargs)
task.start()
return task
def test_uniprot_info(self):
fasta_file = self.setup_tempfile(decorin)
glycan_file = self.setup_tempfile(o_glycans)
db_file = self.setup_tempfile("")
glycan_builder = TextFileGlycanHypothesisSerializer(
glycan_file, db_file)
glycan_builder.start()
glycopeptide_builder = naive_glycopeptide.MultipleProcessFastaGlycopeptideHypothesisSerializer(
fasta_file,
db_file,
glycan_builder.hypothesis_id,
protease=['trypsin'],
constant_modifications=constant_modifications,
variable_modifications=[],
max_missed_cleavages=2)
glycopeptide_builder.start()
post_cleavage = glycopeptide_builder.query(serialize.Peptide).filter(
serialize.Peptide.base_peptide_sequence ==
"DEASGIGPEEHFPEVPEIEPMGPVCPFR").first()
self.assertIsNotNone(post_cleavage)
self.clear_file(db_file)
self.clear_file(fasta_file)
self.clear_file(glycan_file)
def test_missing_glycopeptide(self):
glycan_file = self.setup_tempfile(o_glycans)
fasta_file = self.setup_tempfile(HEPARANASE)
db_file = fasta_file + '.db'
glycan_builder = TextFileGlycanHypothesisSerializer(
glycan_file, db_file)
glycan_builder.start()
glycopeptide_builder = naive_glycopeptide.FastaGlycopeptideHypothesisSerializer(
fasta_file,
db_file,
glycan_builder.hypothesis_id,
protease='trypsin',
constant_modifications=constant_modifications,
max_missed_cleavages=2)
glycopeptide_builder.start()
case = glycopeptide_builder.query(
Glycopeptide.glycopeptide_sequence ==
"KFKNSTYS(O-Glycosylation)R{Hex:1; HexNAc:1; Neu5Ac:2}").first()
self.assertIsNotNone(case)
self.clear_file(glycan_file)
self.clear_file(fasta_file)
self.clear_file(db_file)
def build(self, database_connection, **kwargs):
if kwargs.get('hypothesis_name') is None:
kwargs['hypothesis_name'] = (self.hypothesis_metadata['name'])
task = TextFileGlycanHypothesisSerializer(load_text(),
database_connection,
**kwargs)
task.start()
return task
def build(self, database_connection, **kwargs):
if kwargs.get('hypothesis_name') is None:
kwargs['hypothesis_name'] = (self.hypothesis_metadata['name'])
task = TextFileGlycanHypothesisSerializer(
load_text(), database_connection,
**kwargs)
task.start()
return task
def _build_glycan_composition_hypothesis_from_text(database_connection, text_file, hypothesis_name,
identifier=None):
if hypothesis_name is not None:
hypothesis_name = hypothesis_name + '-Glycans'
builder = TextFileGlycanHypothesisSerializer(
text_file, database_connection, hypothesis_name=hypothesis_name)
builder.run()
return builder.hypothesis_id
def _build_glycan_composition_hypothesis_from_text(database_connection,
text_file,
hypothesis_name,
identifier=None):
if hypothesis_name is not None:
hypothesis_name = hypothesis_name + '-Glycans'
builder = TextFileGlycanHypothesisSerializer(
text_file, database_connection, hypothesis_name=hypothesis_name)
builder.run()
return builder.hypothesis_id
def glycan_text(context, text_file, database_connection, reduction, derivatization, name):
if name is not None:
name = validate_glycan_hypothesis_name(context, database_connection, name)
click.secho("Building Glycan Hypothesis %s" % name, fg='cyan')
reduction = validate_reduction(context, reduction)
derivatization = validate_derivatization(context, derivatization)
builder = TextFileGlycanHypothesisSerializer(
text_file, database_connection, reduction=reduction, derivatization=derivatization,
hypothesis_name=name)
builder.display_header()
builder.start()
def test_throughput(self):
fasta_file = fixtures.get_test_data("phil-82-proteins.fasta")
glycan_file = fixtures.get_test_data("IAV_matched_glycans.txt")
db_file = self.setup_tempfile("")
print(db_file)
glycan_builder = TextFileGlycanHypothesisSerializer(glycan_file, db_file)
glycan_builder.start()
glycopeptide_builder = naive_glycopeptide.MultipleProcessFastaGlycopeptideHypothesisSerializer(
fasta_file, db_file, glycan_builder.hypothesis_id, constant_modifications=constant_modifications,
variable_modifications=variable_modifications, max_missed_cleavages=2)
glycopeptide_builder.start()
self.clear_file(db_file)
def test_extract_forward_backward(self):
fasta_file = fixtures.get_test_data("yeast_glycoproteins.fa")
glycan_file = self.setup_tempfile(simple_n_glycans)
forward_db = self.setup_tempfile("")
reverse_db = self.setup_tempfile("")
glycan_builder = TextFileGlycanHypothesisSerializer(glycan_file, forward_db)
glycan_builder.start()
builder = naive_glycopeptide.MultipleProcessFastaGlycopeptideHypothesisSerializer(
fasta_file, forward_db, 1)
cnt = builder.extract_proteins()
assert cnt == 251
glycan_builder = TextFileGlycanHypothesisSerializer(glycan_file, reverse_db)
glycan_builder.start()
rev_builder = naive_glycopeptide.ReversingMultipleProcessFastaGlycopeptideHypothesisSerializer(
fasta_file, reverse_db, 1)
cnt = rev_builder.extract_proteins()
assert cnt == 251
fwd_prots = builder.query(serialize.Protein).all()
rev_prots = rev_builder.query(serialize.Protein).all()
for fx, rx in zip(fwd_prots, rev_prots):
assert fx.name == rx.name
assert len(fx.glycosylation_sites) == len(rx.glycosylation_sites)
def test_throughput(self):
fasta_file = fixtures.get_test_data("phil-82-proteins.fasta")
glycan_file = fixtures.get_test_data("IAV_matched_glycans.txt")
db_file = self.setup_tempfile("")
print(db_file)
glycan_builder = TextFileGlycanHypothesisSerializer(glycan_file, db_file)
glycan_builder.start()
glycopeptide_builder = naive_glycopeptide.MultipleProcessFastaGlycopeptideHypothesisSerializer(
fasta_file, db_file, glycan_builder.hypothesis_id, constant_modifications=constant_modifications,
variable_modifications=variable_modifications, max_missed_cleavages=2)
glycopeptide_builder.start()
self.clear_file(db_file)
def test_uniprot_info(self):
fasta_file = self.setup_tempfile(decorin)
glycan_file = self.setup_tempfile(o_glycans)
db_file = self.setup_tempfile("")
glycan_builder = TextFileGlycanHypothesisSerializer(glycan_file, db_file)
glycan_builder.start()
glycopeptide_builder = naive_glycopeptide.MultipleProcessFastaGlycopeptideHypothesisSerializer(
fasta_file, db_file, glycan_builder.hypothesis_id,
protease=['trypsin'],
constant_modifications=constant_modifications,
variable_modifications=[], max_missed_cleavages=2)
glycopeptide_builder.start()
post_cleavage = glycopeptide_builder.query(serialize.Peptide).filter(
serialize.Peptide.base_peptide_sequence == "DEASGIGPEEHFPEVPEIEPMGPVCPFR").first()
self.assertIsNotNone(post_cleavage)
self.clear_file(db_file)
self.clear_file(fasta_file)
self.clear_file(glycan_file)
def test_missing_glycopeptide(self):
glycan_file = self.setup_tempfile(o_glycans)
fasta_file = self.setup_tempfile(HEPARANASE)
db_file = fasta_file + '.db'
glycan_builder = TextFileGlycanHypothesisSerializer(glycan_file, db_file)
glycan_builder.start()
glycopeptide_builder = naive_glycopeptide.FastaGlycopeptideHypothesisSerializer(
fasta_file, db_file, glycan_builder.hypothesis_id, protease='trypsin',
constant_modifications=constant_modifications,
max_missed_cleavages=2)
glycopeptide_builder.start()
case = glycopeptide_builder.query(
Glycopeptide.glycopeptide_sequence == "KFKNSTYS(O-Glycosylation)R{Hex:1; HexNAc:1; Neu5Ac:2}").first()
self.assertIsNotNone(case)
self.clear_file(glycan_file)
self.clear_file(fasta_file)
self.clear_file(db_file)
def test_extract_forward_backward(self):
fasta_file = fixtures.get_test_data("yeast_glycoproteins.fa")
glycan_file = self.setup_tempfile(simple_n_glycans)
forward_db = self.setup_tempfile("")
reverse_db = self.setup_tempfile("")
glycan_builder = TextFileGlycanHypothesisSerializer(glycan_file, forward_db)
glycan_builder.start()
builder = naive_glycopeptide.MultipleProcessFastaGlycopeptideHypothesisSerializer(
fasta_file, forward_db, 1)
cnt = builder.extract_proteins()
assert cnt == 251
glycan_builder = TextFileGlycanHypothesisSerializer(glycan_file, reverse_db)
glycan_builder.start()
rev_builder = naive_glycopeptide.ReversingMultipleProcessFastaGlycopeptideHypothesisSerializer(
fasta_file, reverse_db, 1)
cnt = rev_builder.extract_proteins()
assert cnt == 251
fwd_prots = builder.query(serialize.Protein).all()
rev_prots = rev_builder.query(serialize.Protein).all()
for fx, rx in zip(fwd_prots, rev_prots):
assert fx.name == rx.name
assert len(fx.glycosylation_sites) == len(rx.glycosylation_sites)
def glycan_text(context, text_file, database_connection, reduction,
derivatization, name):
if name is not None:
name = validate_glycan_hypothesis_name(context, database_connection,
name)
click.secho("Building Glycan Hypothesis %s" % name, fg='cyan')
reduction = validate_reduction(context, reduction)
derivatization = validate_derivatization(context, derivatization)
builder = TextFileGlycanHypothesisSerializer(text_file,
database_connection,
reduction=reduction,
derivatization=derivatization,
hypothesis_name=name)
builder.display_header()
builder.start()