def test_parallel_database_mapper_usearch(self):
""" parallel_database_mapper_usearch functions as expected """
params = {
'refseqs_fp': self.refseqs1_fp,
'min_percent_id': 0.97,
'evalue': 1e-10,
'max_accepts': 1,
'max_rejects': 32,
'queryalnfract': 0.35,
'targetalnfract': 0.0,
'observation_metadata_fp': None
}
app = ParallelDatabaseMapperUsearch()
r = app(self.inseqs1_fp,
self.test_out,
params,
job_prefix='PTEST',
poll_directly=True,
suppress_submit_jobs=False)
observation_map_fp = glob(join(self.test_out,
'observation_map.txt'))[0]
omap = parse_otu_map(open(observation_map_fp, 'U'))
self.assertEqual(len(omap[0]), 3)
self.assertItemsEqual(
omap[1], ['eco:b0015', 'eco:b0122', 'eco:b0015:duplicate'])
self.assertItemsEqual(omap[2], ['eco:b0015-pr', 'eco:b0122-pr'])
def make_otu_table(otu_map_f,
otu_to_taxonomy=None,
delim='_',
table_id=None,
sample_metadata=None,
constructor=SparseOTUTable):
data, sample_ids, otu_ids = parse_otu_map(otu_map_f,delim)
if otu_to_taxonomy != None:
otu_metadata = []
for o in otu_ids:
try:
otu_metadata.append({'taxonomy':otu_to_taxonomy[o].split(';')})
except KeyError:
otu_metadata.append({'taxonomy':["None"]})
else:
otu_metadata = None
if sample_metadata != None:
raise NotImplementedError,\
"Passing of sample metadata to make_otu_table is not currently supported."
try:
otu_table = table_factory(data, sample_ids, otu_ids,
sample_metadata=sample_metadata,
observation_metadata=otu_metadata,
table_id=table_id,
constructor=constructor,
dtype=int)
except ValueError,e:
raise ValueError,\
("Couldn't create OTU table. Is your OTU map empty?"
" Original error message: %s" % (str(e)))
def test_parallel_database_mapper_usearch(self):
""" parallel_database_mapper_usearch functions as expected """
params = {'refseqs_fp': self.refseqs1_fp,
'min_percent_id': 0.97,
'evalue': 1e-10,
'max_accepts': 1,
'max_rejects': 32,
'queryalnfract': 0.35,
'targetalnfract': 0.0,
'observation_metadata_fp': None
}
app = ParallelDatabaseMapperUsearch()
r = app(self.inseqs1_fp,
self.test_out,
params,
job_prefix='PTEST',
poll_directly=True,
suppress_submit_jobs=False)
observation_map_fp = glob(
join(self.test_out, 'observation_map.txt'))[0]
omap = parse_otu_map(open(observation_map_fp, 'U'))
self.assertEqual(len(omap[0]), 3)
self.assertItemsEqual(
omap[1],
['eco:b0015',
'eco:b0122',
'eco:b0015:duplicate'])
self.assertItemsEqual(omap[2], ['eco:b0015-pr', 'eco:b0122-pr'])
def make_otu_table(otu_map_f, otu_to_taxonomy=None, delim='_', table_id=None,
otu_ids_to_exclude=None, sample_metadata=None, seq_counts=None):
"""Generate a BIOM table from an OTU map
Parameters
----------
otu_map_f : file-like object
The OTU map. Jagged tab-separated file where the first column contains
the OTU ID and subsequent columns contain sequence IDs belonging to
that OTU
otu_to_taxonomy : dict, optional
Defaults to ``None``. If supplied, the dict maps OTU IDs to taxonomies
delim : str, optional
Defaults to "_". The delimiter that is used in the sequence IDs to join
the sample ID to the sequence number
table_id : object, optional
Defaults to ``None``. The identifier that will be given to the
generated BIOM table
otu_ids_to_exclude : iterable, optional
Defaults to ``None``. If present, these OTUs will not be added to the
OTU table from the OTU map
sample_metadata : dict of dicts, optional
Defaults to ``None``. If supplied, keys in the outer dict should be
sample IDs, and keys in the inner dicts should be column names.
seq_counts : dict, optional
Defaults to ``None``. If supplied, the dict maps seq ids to seq counts
"""
data, sample_ids, otu_ids = parse_otu_map(
otu_map_f, delim=delim, otu_ids_to_exclude=otu_ids_to_exclude,seq_counts=seq_counts)
if otu_to_taxonomy is not None:
otu_metadata = []
for o in otu_ids:
otu_metadata.append({'taxonomy': otu_to_taxonomy.get(o, ["None"])})
else:
otu_metadata = None
# if sample_metadata is supplied, put in index-order with the OTU map's
# sample_ids, and do not include samples that were in the mapping file
# but NOT in the OTU map
if sample_metadata is not None:
try:
sample_metadata = [sample_metadata[sample_id]
for sample_id in sample_ids]
except KeyError:
raise KeyError("Sample IDs found in OTU map without sample "
"metadata")
try:
return Table(data, otu_ids, sample_ids,
observation_metadata=otu_metadata,
sample_metadata=sample_metadata, table_id=table_id,
generated_by=get_generated_by_for_biom_tables(),
create_date=datetime.now().isoformat())
except ValueError as e:
raise ValueError("Couldn't create OTU table. Is your OTU map empty?"
" Original error message: %s" % (str(e)))
def test_find_otus(self):
""" test_find_otus: This fxn checks for otu assignment of a set of
sequences
"""
#make sure files gets cleaned up
self.files_to_remove.append(self.output_otu_map)
self.files_to_remove.append(self.leftovers)
self.files_to_remove.append(self.input_fasta)
#run the fxn
find_otus(self.input_fasta,self.leftovers,self.output_otu_map)
data, sample_ids, otu_ids = parse_otu_map(open(self.output_otu_map))
#check the outputs are correct
self.assertEqual(sample_ids[0],'HIT')
self.assertEqual(open(self.leftovers).read(),exp_failures)
def make_otu_table(otu_map_f, otu_to_taxonomy=None, delim='_', table_id=None,
otu_ids_to_exclude=None, sample_metadata=None):
"""Generate a BIOM table from an OTU map
Parameters
----------
otu_map_f : file-like object
The OTU map. Jagged tab-separated file where the first column contains
the OTU ID and subsequent columns contain sequence IDs belonging to
that OTU
otu_to_taxonomy : dict, optional
Defaults to ``None``. If supplied, the dict maps OTU IDs to taxonomies
delim : str, optional
Defaults to "_". The delimiter that is used in the sequence IDs to join
the sample ID to the sequence number
table_id : object, optional
Defaults to ``None``. The identifier that will be given to the
generated BIOM table
otu_ids_to_exclude : iterable, optional
Defaults to ``None``. If present, these OTUs will not be added to the
OTU table from the OTU map
sample_metadata : iterable of dicts, optional
Defaults to ``None``.
"""
data, sample_ids, otu_ids = parse_otu_map(
otu_map_f, delim=delim, otu_ids_to_exclude=otu_ids_to_exclude)
if otu_to_taxonomy is not None:
otu_metadata = []
for o in otu_ids:
otu_metadata.append({'taxonomy': otu_to_taxonomy.get(o, ["None"])})
else:
otu_metadata = None
# if sample_metadata is not None:
# raise NotImplementedError("Passing of sample metadata to "
# "make_otu_table is not currently supported.")
try:
return Table(data, otu_ids, sample_ids,
observation_metadata=otu_metadata,
sample_metadata=sample_metadata, table_id=table_id,
generated_by=get_generated_by_for_biom_tables(),
create_date=datetime.now().isoformat())
except ValueError as e:
raise ValueError("Couldn't create OTU table. Is your OTU map empty?"
" Original error message: %s" % (str(e)))
def test_parallel_database_mapper_usearch(self):
""" parallel_database_mapper_usearch functions as expected """
params = {
"refseqs_fp": self.refseqs1_fp,
"min_percent_id": 0.97,
"evalue": 1e-10,
"max_accepts": 1,
"max_rejects": 32,
"queryalnfract": 0.35,
"targetalnfract": 0.0,
"observation_metadata_fp": None,
}
app = ParallelDatabaseMapperUsearch()
r = app(
self.inseqs1_fp, self.test_out, params, job_prefix="PTEST", poll_directly=True, suppress_submit_jobs=False
)
observation_map_fp = glob(join(self.test_out, "observation_map.txt"))[0]
omap = parse_otu_map(open(observation_map_fp, "U"))
self.assertEqual(len(omap[0]), 3)
self.assertItemsEqual(omap[1], ["eco:b0015", "eco:b0122", "eco:b0015:duplicate"])
self.assertItemsEqual(omap[2], ["eco:b0015-pr", "eco:b0122-pr"])
def test_parallel_function_assigner_usearch(self):
""" parallel_function_assigner_usearch functions as expected """
params = {'refseqs_fp':self.refseqs1_fp,
'min_percent_id':0.97,
'evalue':1e-10,
'max_accepts':1,
'max_rejects':32,
'queryalnfract':0.35,
'targetalnfract':0.0
}
app = ParallelFunctionAssignerUsearch()
r = app(self.inseqs1_fp,
self.test_out,
params,
job_prefix='PTEST',
poll_directly=True,
suppress_submit_jobs=False)
otu_map_fp = glob(join(self.test_out,'*fmap.txt'))[0]
fmap = parse_otu_map(open(otu_map_fp,'U'))
self.assertEqual(len(fmap[0]),2)
self.assertEqualItems(fmap[1],['eco:b0015', 'eco:b0122'])
self.assertEqualItems(fmap[2],['eco:b0015-pr', 'eco:b0122-pr'])
def make_otu_table(otu_map_f,
otu_to_taxonomy=None,
delim='_',
table_id=None,
otu_ids_to_exclude=None,
sample_metadata=None):
"""Generate a BIOM table from an OTU map
Parameters
----------
otu_map_f : file-like object
The OTU map. Jagged tab-separated file where the first column contains
the OTU ID and subsequent columns contain sequence IDs belonging to
that OTU
otu_to_taxonomy : dict, optional
Defaults to ``None``. If supplied, the dict maps OTU IDs to taxonomies
delim : str, optional
Defaults to "_". The delimiter that is used in the sequence IDs to join
the sample ID to the sequence number
table_id : object, optional
Defaults to ``None``. The identifier that will be given to the
generated BIOM table
otu_ids_to_exclude : iterable, optional
Defaults to ``None``. If present, these OTUs will not be added to the
OTU table from the OTU map
sample_metadata : dict of dicts, optional
Defaults to ``None``. If supplied, keys in the outer dict should be
sample IDs, and keys in the inner dicts should be column names.
"""
data, sample_ids, otu_ids = parse_otu_map(
otu_map_f, delim=delim, otu_ids_to_exclude=otu_ids_to_exclude)
if otu_to_taxonomy is not None:
otu_metadata = []
for o in otu_ids:
otu_metadata.append({'taxonomy': otu_to_taxonomy.get(o, ["None"])})
else:
otu_metadata = None
# if sample_metadata is supplied, put in index-order with the OTU map's
# sample_ids, and do not include samples that were in the mapping file
# but NOT in the OTU map
if sample_metadata is not None:
try:
sample_metadata = [
sample_metadata[sample_id] for sample_id in sample_ids
]
except KeyError:
raise KeyError("Sample IDs found in OTU map without sample "
"metadata")
try:
return Table(data,
otu_ids,
sample_ids,
observation_metadata=otu_metadata,
sample_metadata=sample_metadata,
table_id=table_id,
generated_by=get_generated_by_for_biom_tables(),
create_date=datetime.now().isoformat())
except ValueError as e:
raise ValueError("Couldn't create OTU table. Is your OTU map empty?"
" Original error message: %s" % (str(e)))
