def main():
option_parser, opts, args = parse_command_line_parameters(**script_info)
if not (opts.refseqs_path or opts.blast_db):
option_parser.error('Either a blast db (via -b) or a collection of '
'reference sequences (via -r) must be passed')
if opts.refseqs_path and opts.blast_db:
option_parser.error('You should provide only a blast db (via -b) '
'or a collection of reference sequences (via -r), but not both')
# create dict of command-line options
params = eval(str(opts))
parallel_runner = ParallelBlaster(
cluster_jobs_fp=opts.cluster_jobs_fp,
jobs_to_start=opts.jobs_to_start,
retain_temp_files=opts.retain_temp_files,
suppress_polling=opts.suppress_polling,
seconds_to_sleep=opts.seconds_to_sleep)
parallel_runner(opts.infile_path,
opts.output_dir,
params,
job_prefix=opts.job_prefix,
poll_directly=opts.poll_directly,
suppress_submit_jobs=False)
def test_parallel_blaster(self):
"""Test ParallelBlaster functions as expected."""
params = {
'refseqs_path': self.reference_seqs_file.name,
'disable_low_complexity_filter': False,
'e_value': 0.001,
'num_hits': 1,
'word_size': 30,
'suppress_format_blastdb': False,
'blastmat_dir': None
}
app = ParallelBlaster()
r = app(self.tmp_seq_filepath,
self.test_out,
params,
job_prefix='BLASTTEST',
poll_directly=True,
suppress_submit_jobs=False)
# Basic sanity checks: we should get two blast hits (lines). We ignore
# all of the comments in the file. Each line should have 12 fields
# separated by tabs.
results = [
line for line in open(
glob(join(self.test_out, '*_blast_out.txt'))[0], 'U')
if not line.startswith('#')
]
self.assertEqual(len(results), 2)
self.assertEqual(len(results[0].split('\t')), 12)
self.assertEqual(len(results[1].split('\t')), 12)
def main():
option_parser, opts, args = parse_command_line_parameters(**script_info)
if not ((exists(opts.refseqs_path) and isfile(opts.refseqs_path)) or \
(opts.suppress_format_blastdb and
len(glob('%s*nin' % opts.refseqs_path)) > 0)):
option_parser.error('%s (-r) doesn\'t exist as a fasta file or '
'pre-formatted BLAST database. If passing a '
'reference database you must also pass -D.' %
opts.refseqs_path)
# create dict of command-line options
params = eval(str(opts))
parallel_runner = ParallelBlaster(
cluster_jobs_fp=opts.cluster_jobs_fp,
jobs_to_start=opts.jobs_to_start,
retain_temp_files=opts.retain_temp_files,
suppress_polling=opts.suppress_polling,
seconds_to_sleep=opts.seconds_to_sleep)
parallel_runner(opts.infile_path,
opts.output_dir,
params,
job_prefix=opts.job_prefix,
poll_directly=opts.poll_directly,
suppress_submit_jobs=False)