def has_trait(self, trait):
h5files = fsutils.get_h5files_in_dir(self.search_path, self.trait_dir)
for h5file in h5files:
service = trait_service.TraitService(h5file=h5file)
if service.has_trait(trait):
return True
raise NotFoundError("Trait " + trait)
def _location_for_snp_in_chromosome(self, chromosome):
"""
Looks up all the files under the chromosome directory, in parallel, and tries to find out which one the
SNP lives in.
:param chromosome: the chromosome we think the SNP lives in
:return: the exact file that this SNP is stored in or raises and error if not found
"""
dir_name = fsutils.join(self.snp_dir, str(chromosome))
if not fsutils.is_valid_dir_path(path=self.search_path,
dir_name=dir_name):
logger.debug("Chromosome %s given for variant %s doesn't exist!",
str(self.chromosome), self.snp)
raise NotFoundError("Chromosome " + str(self.chromosome))
h5files = fsutils.get_h5files_in_dir(path=self.search_path,
dir_name=dir_name)
snps = [self.snp for _ in h5files]
pool = Pool(self.bp_step)
results = pool.map(is_snp_in_file, zip(snps, h5files))
pool.close()
pool.join()
for h5file in results:
if h5file is not None:
return h5file
# not found anywhere in chromosome
raise NotFoundError("Chromosome-variant combination")
def get_list_of_studies(self):
studies = []
h5files = fsutils.get_h5files_in_dir(self.search_path, self.trait_dir)
for h5file in h5files:
service = study_service.StudyService(h5file=h5file)
studies.extend(service.list_studies())
service.close_file()
return sorted(studies)
def get_list_of_traits(self):
traits = []
h5files = fsutils.get_h5files_in_dir(self.search_path, self.trait_dir)
for h5file in h5files:
service = trait_service.TraitService(h5file=h5file)
traits.extend(service.list_traits())
service.close_file()
return sorted(traits)
def get_trait_of_study(self, study_to_find):
h5files = fsutils.get_h5files_in_dir(self.search_path, self.trait_dir)
for h5file in h5files:
service = study_service.StudyService(h5file=h5file)
for trait_study in service.list_trait_study_pairs():
if study_to_find == trait_study.split(":")[1]:
service.close_file()
return trait_study.split(":")[0]
service.close_file()
# study not found
raise NotFoundError("Study " + study_to_find)
def has_chromosome(self, chromosome):
# raises Not Found Error
"""To do: Store the chromosome list as an attribute in the hdf5 file."""
h5files = fsutils.get_h5files_in_dir(self.search_path, self.study_dir)
#chromosomes = []
#for h5file in h5files:
# service = trait_service.StudyService(h5file=h5file)
# traits.extend(service.list_traits())
# service.close_file()
search = cr.search_all_assocs(chromosome=chromosome,
start=0,
size=0,
properties=self.properties)
if search[-1] > 0:
print('checked')
return True
raise NotFoundError("Chromosome " + str(chromosome))
def _find_h5file_study_group(self):
"""
Traverse all the hdf5 file and find any with the study group of interest
:return: dict of {h5file: studygroup path}
"""
hf_study_dict = {}
h5files = fsutils.get_h5files_in_dir(self.search_path, self.trait_dir)
for h5file in h5files:
service = study_service.StudyService(h5file=h5file)
for study_group in service.get_study_groups():
if self.study == study_group.get_name().split("/")[-1]:
hf_study_dict[h5file] = study_group.get_name()
service.close_file()
if any(hf_study_dict):
return hf_study_dict
else:
logger.debug("Study %s not found in any trait!", self.study)
raise NotFoundError("Study " + self.study)
def test_get_h5files_in_dir_raise_error(self):
with pytest.raises(RuntimeError):
fsu.get_h5files_in_dir('/', self.dir_name)
def test_get_h5files_in_dir_same_order(self):
files1 = fsu.get_h5files_in_dir('./', self.dir_name)
files2 = fsu.get_h5files_in_dir('./', self.dir_name)
assert np.array_equal(files1, files2)