def __init__(self, dataset):
self.dataset = dataset
folder_path = PathResolver.output_path_for(self.MAPPINGS_FOLDER)
PathResolver.assure_path_exists(folder_path)
db_path = PathResolver.output_path_for(self.MAPPINGS_FOLDER,
self.DB_FOLDER)
PathResolver.assure_path_exists(db_path)
def perform(self):
PathResolver.assure_path_exists(self._db_folder_path())
all_vs_all_path = PathResolver.output_path_for(
PathResolver.ALL_VS_ALL_FOLDER)
PathResolver.assure_path_exists(all_vs_all_path)
one_vs_all_path = PathResolver.output_path_for(
PathResolver.ONE_VS_ALL_FOLDER)
PathResolver.assure_path_exists(one_vs_all_path)
self._make_databases()
self._perform_blasts()
self._make_one_vs_all_files()
def _make_one_vs_all_files(self):
all_vs_all_path = PathResolver.output_path_for(
PathResolver.ALL_VS_ALL_FOLDER)
files = {}
for f_path in glob.glob(all_vs_all_path +
'/*.%s' % self.BLAST_RESULT_EXT):
left_org_id = os.path.splitext(
os.path.basename(f_path))[0].split('_VS_')[0]
if left_org_id not in files:
files[left_org_id] = []
files[left_org_id].append(f_path)
for org_id, files in files.iteritems():
org_file_name = org_id + '.' + self.BLAST_RESULT_EXT
path = PathResolver.output_path_for(PathResolver.ONE_VS_ALL_FOLDER,
org_file_name)
if os.path.exists(path):
os.remove(path)
for f_path in files:
subprocess.call('cat %s >> %s' % (f_path, path), shell=True)
def _perform_blast(self, left_path, right_path):
command = 'blastn -query %s -db %s -out %s -outfmt "6 %s" -num_threads %s'
right_org_name = os.path.splitext(os.path.basename(right_path))[0]
left_org_name = os.path.splitext(os.path.basename(left_path))[0]
db_path = self._db_folder_path(right_org_name)
outfile_name = '%s_VS_%s.%s' % (left_org_name, right_org_name,
self.BLAST_RESULT_EXT)
output_path = PathResolver.output_path_for(
PathResolver.ALL_VS_ALL_FOLDER, outfile_name)
threads_cnt = Settings.winston.tools.blast.threads
command = command % (left_path, db_path, output_path, self.COLUMNS,
threads_cnt)
subprocess.call(command, shell=True)
def db_path(self):
return PathResolver.output_path_for(self.MAPPINGS_FOLDER,
self.DB_FOLDER)
def pileup_output_path(self):
file_name = '%s_pileup.txt' % self.dataset.external_name
return PathResolver.output_path_for(self.MAPPINGS_FOLDER, file_name)
def sam_file_path(self):
file_name = '%s.sam' % self.dataset.external_name
paths = [self.MAPPINGS_FOLDER, file_name]
return PathResolver.output_path_for(*paths)
def _dict_file_path():
return PathResolver.output_path_for(NameConverter.DICT_FILENAME)
def pair_types_path():
return PathResolver.output_path_for(PathResolver.TYPES_FILENAME)
def _db_folder_path(self, *inner_path):
return PathResolver.output_path_for(PathResolver.BLAST_DB_FOLDER,
*inner_path)
def _default_db_path():
return PathResolver.output_path_for(
DatabaseWorker.DEFAULT_DB_FILE_NAME)