def test_4_query_custom_small_compressed(self):
""" Query gzip compressed fastq files (less than 20MB).
"""
for sample in glob.glob(os.path.join(self.custom_dir, "*.fastq.gz")):
print "\nQuerying against compressed sample %s" % sample
if os.path.getsize(os.path.join(self.custom_dir, sample)) < 20*1024*1204:
facs.query(os.path.join(self.custom_dir, sample),os.path.join(self.bloom_dir, "U00096.2.bloom"))
def test_3_query_custom(self):
""" Query against the uncompressed FastQ files files manually deposited in data/custom folder.
"""
for sample in glob.glob(os.path.join(self.custom_dir, "*.fastq")):
print "\nQuerying against uncompressed sample %s" % sample
facs.query(os.path.join(self.custom_dir, sample),
os.path.join(self.bloom_dir, "U00096.2.bloom"))
def test_2_query(self):
""" Generate dummy fastq files.
"""
for nreads in self.fastq_nreads:
test_fname = "test%s.fastq" % nreads
helper._generate_dummy_fastq(os.path.join(self.synthetic_fastq, test_fname), nreads)
facs.query(os.path.join(self.synthetic_fastq, test_fname),
os.path.join(self.bloom_dir, "U00096.2.bloom"))
def test_query_NA21137_illumina(self):
""" Query gzip compressed fastq files
"""
for sample in glob.glob(os.path.join(self.custom_dir, "*.fastq.gz")):
sample_path = os.path.join(self.custom_dir, sample)
for ref in os.listdir(self.reference):
bf = os.path.join(self.bloom_dir, os.path.splitext(ref)[0]+".bloom")
print(sample_path, bf)
facs.query(sample_path, bf)
def test_3_query_custom(self):
""" Query against the uncompressed FastQ files files manually deposited
in data/custom folder.
"""
for sample in glob.glob(os.path.join(self.custom_dir, "*.fastq")):
print "\nQuerying against uncompressed sample %s" % sample
for ref in os.listdir(self.reference):
facs.query(os.path.join(self.synthetic_fastq, test_fname),
os.path.join(self.bloom_dir, os.path.splitext(ref)[0]+".bloom"))
def test_2_query_exampledata(self):
""" Query Innocentive exampledata
"""
for sample in glob.glob(os.path.join(self.example, "*.fq")):
print "\nQuerying against example dataset %s" % sample
for ref in os.listdir(self.reference):
example = os.path.join(self.example, sample)
bf = os.path.join(self.bloom_dir, os.path.splitext(ref)[0]+".bloom")
print(example, bf)
facs.query(example, bf)
def test_2_query_testdata(self):
""" Query Innocentive testdata
"""
for sample in glob.glob(os.path.join(self.testing, "*.fq")):
print "\nQuerying against testing dataset %s" % sample
for ref in os.listdir(self.reference):
testset = os.path.join(self.testing, sample)
bf = os.path.join(self.bloom_dir, os.path.splitext(ref)[0]+".bloom")
print(testset, bf)
facs.query(testset, bf)
def test_2_query(self):
""" Generate dummy fastq files and query them against reference bloom filters.
"""
for nreads in self.fastq_nreads:
test_fname = "test%s.fastq" % nreads
helpers.generate_dummy_fastq(os.path.join(self.synthetic_fastq,
test_fname), nreads)
for ref in os.listdir(self.reference):
qry = os.path.join(self.synthetic_fastq, test_fname)
bf = os.path.join(self.bloom_dir, os.path.splitext(ref)[0]+".bloom")
print(qry, bf)
facs.query(qry, bf)
def test_2_query(self):
""" Query against datasets (generated by simNGS)
"""
for sample in glob.glob(os.path.join(self.synthetic_fastq, "*.fastq")):
for ref in os.listdir(self.reference):
qry = os.path.join(self.synthetic_fastq, sample)
bf = os.path.join(self.bloom_dir, os.path.splitext(ref)[0]+".bloom")
print(qry, bf)
json_doc = facs.query(qry, bf)
self.results.append(json_doc)
def test_2_query(self):
""" Query against datasets (generated by simNGS)
"""
for sample in glob.glob(os.path.join(self.synthetic_fastq, "*.fastq")):
for ref in os.listdir(self.reference):
qry = os.path.join(self.synthetic_fastq, sample)
bf = os.path.join(self.bloom_dir, os.path.splitext(ref)[0]+".bloom")
print(qry, bf)
mem = [-1]
if profile:
mem, json_doc = memory_usage((facs.query, (qry, bf),), include_children=True, retval=True)
else:
json_doc = facs.query(qry, bf)
json_dict = json.loads(json_doc)
json_dict['memory_usage'] = mem
self.results.append(json.dumps(json_dict))
def test_2_run_low_complexity(self):
#XXX collate the JSON output in a better way
for sample in glob.glob(os.path.join(self.synthetic, "*.fna")):
for ref in os.listdir(self.reference):
facs.query(os.path.join(self.synthetic, sample),
os.path.join(self.bloom_dir, os.path.splitext(ref)[0]+".bloom"))
def test_query_NA21137_illumina(self):
""" Query gzip compressed fastq files
"""
for sample in glob.glob(os.path.join(self.custom_dir, "*.fastq.gz")):
facs.query(os.path.join(self.custom_dir, sample), os.path.join(self.bloom_dir, "U00096.2.bloom"))