def set_parameters(self, args):
"""Validate and set parameters.
Parameters
----------
args : dict
command-line arguments
Raises
------
ValueError
found invalid parameter(s)
"""
# load arguments
for key, val in vars(args).items():
setattr(self, key, val)
# check input directory and data
if self.input:
if isfile(self.input):
self.input_map = {file2id(self.input): self.input}
elif isdir(self.input):
self.input_map = {
k: join(self.input, v)
for k, v in id2file_map(self.input, ext='tsv').items()
}
else:
raise ValueError(
'Invalid input data file or directory: {}.'.format(
self.input))
if len(self.input_map) == 0:
raise ValueError('No input data are found under: {}.'.format(
self.input))
# check / create output directory
makedirs(self.output, exist_ok=True)
self.prev_map = id2file_map(self.output, 'tsv')
# load configurations
get_config(self, 'evalue', 'analyze.evalue', float)
for key in ('maxhits', 'identity', 'coverage'):
get_config(self, key, 'analyze.{}'.format(key))
for key in ('input_cov', 'self_rank', 'close_size', 'distal_top'):
get_config(self, key, 'grouping.{}'.format(key.replace('_', '')))
for key in ('weighted', 'outliers', 'orphans', 'bandwidth', 'bw_steps',
'low_part', 'noise', 'fixed', 'silhouette', 'self_low'):
get_config(self, key, 'predict.{}'.format(key.replace('_', '')))
# convert boolean values
for key in ('weighted', 'orphans', 'self_low'):
setattr(self, key, arg2bool(getattr(self, key, None)))
# convert fractions to percentages
for metric in ('input_cov', 'noise', 'fixed', 'distal_top'):
val = getattr(self, metric)
if val and val < 1:
setattr(self, metric, val * 100)
# convert distal top to a factor to save compute
self.match_th = 1 - self.distal_top / 100
# force coverage >= 50 to ensure that candidates are sequential
if (self.input_cov or 0) < 50:
raise ValueError('Taxonomy coverage for auto-interence must be at '
'least 50%.')
def test_file2id(self):
self.assertEqual(file2id('sample.fna'), 'sample')
self.assertEqual(file2id('sample.fna.gz'), 'sample')
self.assertEqual(file2id('/path/to/sample.fna'), 'sample')
self.assertEqual(file2id('sample'), 'sample')
self.assertEqual(file2id(''), '')