def test_prebuild_index(self):
test_dir = tempfile.mkdtemp()
work_path = os.path.join(test_dir, self.path)
with open(work_path, 'w') as dest, open(self.path) as source:
dest.write(source.read())
assert dest.closed
with mgf.IndexedMGF(work_path) as inst:
offsets_exist = os.path.exists(inst._byte_offset_filename)
self.assertEqual(offsets_exist, inst._check_has_byte_offset_file())
self.assertTrue(isinstance(inst._offset_index, aux.OffsetIndex))
self.assertTrue(inst._source.closed)
mgf.IndexedMGF.prebuild_byte_offset_file(work_path)
with mgf.IndexedMGF(work_path) as inst:
offsets_exist = os.path.exists(inst._byte_offset_filename)
self.assertTrue(offsets_exist)
self.assertEqual(offsets_exist, inst._check_has_byte_offset_file())
self.assertTrue(isinstance(inst._offset_index, aux.OffsetIndex))
self.assertTrue(inst._source.closed)
os.remove(inst._byte_offset_filename)
with mgf.IndexedMGF(work_path) as inst:
offsets_exist = os.path.exists(inst._byte_offset_filename)
self.assertEqual(offsets_exist, inst._check_has_byte_offset_file())
self.assertTrue(isinstance(inst._offset_index, aux.OffsetIndex))
self.assertTrue(inst._source.closed)
shutil.rmtree(test_dir, True)
def _create_parser(self):
if self._use_index:
return mgf.IndexedMGF(self.source_file,
read_charges=False,
convert_arrays=1,
encoding=self.encoding)
else:
return mgf.MGF(self.source_file,
read_charges=False,
convert_arrays=1,
encoding=self.encoding)
def process_maracluster_mgf(fname, get_cluster=get_cluster_id,
get_pepmass=naive_average_mass_and_charge,
get_rt=median_rt,
**kwargs):
outputs = []
with mgf.IndexedMGF(fname, read_charges=False) as fin:
titles = fin.index.keys()
for cluster_id, cluster_group in groupby(titles, get_cluster):
ids = list(cluster_group)
spectra = fin[ids]
mz, c = get_pepmass(spectra)
rt = get_rt(spectra)
outputs.append(average_spectrum(
spectra, cluster_id, pepmass=mz, charge=c, rtinseconds=rt, **kwargs))
return outputs
def test_write_index_keys(self):
test_dir = tempfile.mkdtemp()
work_path = os.path.join(test_dir, self.path)
with open(work_path, 'wb') as dest, open(self.path, 'rb') as source:
dest.write(source.read())
assert dest.closed
mgf.IndexedMGF.prebuild_byte_offset_file(work_path)
with mgf.IndexedMGF(work_path) as inst:
ipath = inst._byte_offset_filename
with open(ipath) as ifp:
container = json.load(ifp, object_hook=OrderedDict)
tag_key = mgf.IndexedMGF._index_class._schema_version_tag_key
self.assertEqual(set(container.keys()), {tag_key, 'index'})
self.assertEqual(tuple(container[tag_key]),
mgf.IndexedMGF._index_class.schema_version)
self.assertEqual(
container['index'],
[['Spectrum 1', [217, 343]], ['Spectrum 2', [343, 506]]])
def test_map(self):
with mgf.IndexedMGF(self.path) as reader:
spectra = sorted(list(reader.map()),
key=lambda s: s['params']['title'])
self.assertEqual(data.mgf_spectra_long, spectra)
def test_indexedmgf_picklable(self):
with mgf.IndexedMGF(self.path) as reader:
spec = pickle.dumps(reader)
with pickle.loads(spec) as reader:
self.assertEqual(data.mgf_spectra_long[0], next(reader))
def test_read_list(self):
key = ['Spectrum 2', 'Spectrum 1']
with mgf.IndexedMGF(self.path) as f:
self.assertEqual(data.mgf_spectra_long[::-1], f[key])
def read_mgf(file_path):
return mgf.IndexedMGF(file_path)
def test_key_access_ions(self):
with mgf.IndexedMGF(self.path_annotated, read_ions=True) as f:
np.testing.assert_array_equal(
f['RAEYWENYPPAH||3']['ion array'],
self.spectra_annotated[1]['ion array'])
