def genotype(args):
"""
Genotype structural variants.
:param args: Command arguments.
:return: 0 if the command ran successfully, and a non-zero code otherwise.
"""
print('Genotyping SVs')
return_code = smrtsvrunner.run_snake_target(
'rules/genotype.snakefile', args, PROCESS_ENV, SMRTSV_DIR,
(
'gt_vcf_write',
'--config',
'genotyper_config={}'.format(args.genotyper_config),
'genotyped_variants={}'.format(args.genotyped_variants),
'gt_mapq={}'.format(args.gt_mapq),
'gt_map_cpu={}'.format(args.gt_map_cpu),
'gt_map_mem={}'.format(args.gt_map_mem),
'gt_map_disk={}'.format(args.gt_map_disk),
'gt_map_time={}'.format(args.gt_map_time),
'gt_keep_temp={}'.format(args.gt_keep_temp)
)
)
if return_code != 0:
sys.stderr.write('Failed to genotype SVs\n')
return return_code
def detect(args):
"""
Detect SVs from signatures in read alignments.
"""
print('Detecting variant signatures')
command = (
'detect_group_merge_regions',
'--config',
'assembly_window_size={}'.format(args.assembly_window_size),
'assembly_window_slide={}'.format(args.assembly_window_slide),
'mapping_quality={}'.format(args.mapping_quality),
'min_length={}'.format(args.min_length),
'min_support={}'.format(args.min_support),
'max_support={}'.format(args.max_support),
'min_coverage={}'.format(args.min_coverage),
'max_coverage={}'.format(args.max_coverage),
'min_hardstop_support={}'.format(args.min_hardstop_support),
'max_candidate_length={}'.format(args.max_candidate_length)
)
if args.exclude:
command = command + ('exclude={}'.format(args.exclude),)
return smrtsvrunner.run_snake_target('rules/detect.snakefile', args, PROCESS_ENV, SMRTSV_DIR, command, cmd_log='detect/log')
def assemble(args):
"""
Assemble candidate regions from raw reads aligned to regions.
"""
print('Running local assemblies')
# Quote alignment params
args.asm_alignment_parameters = '"{}"'.format(
args.asm_alignment_parameters)
# Run
command = ('asm_merge_groups', '--config',
'asm_alignment_parameters={}'.format(
args.asm_alignment_parameters),
'mapping_quality={}'.format(args.mapping_quality),
'asm_cpu={}'.format(args.asm_cpu), 'asm_mem={}'.format(
args.asm_mem), 'asm_polish={}'.format(args.asm_polish),
'no_rm_temp={}'.format(args.nt))
return smrtsvrunner.run_snake_target('rules/assemble.snakefile',
args,
PROCESS_ENV,
SMRTSV_DIR,
command,
cmd_log='assemble/log')
def index(args):
print('Preparing reference')
return smrtsvrunner.run_snake_target(
'rules/reference.snakefile',
args,
PROCESS_ENV,
SMRTSV_DIR,
('ref_all', '--config', 'reference={}'.format(args.reference)),
cmd_log='reference/log')
def align(args):
print('Aligning sequence reads')
return smrtsvrunner.run_snake_target(
'rules/align.snakefile',
args,
PROCESS_ENV,
SMRTSV_DIR,
('aln_run', '--config', 'reads={}'.format(args.reads),
'batches={}'.format(args.batches), 'threads={}'.format(args.threads),
'alignment_parameters="{}"'.format(args.alignment_parameters)),
cmd_log='align/log')
def call(args):
# Call SVs, indels, and inversions.
sys.stdout.write("Calling variants\n")
return_code = smrtsvrunner.run_snake_target(
'rules/call.snakefile',
args,
PROCESS_ENV,
SMRTSV_DIR, ('call_variant_vcf', '--config', 'variants={}'.format(
args.variants), 'species="{}"'.format(
args.species), 'sample="{}"'.format(args.sample)),
cmd_log='call/log')
if return_code != 0:
sys.stderr.write('Failed to call variants\n')
return return_code
print('LD_LIBRARY_PATH:')
for PATH_ELEMENT in process_env['LD_LIBRARY_PATH'].split(':'):
print('\t* {}'.format(PATH_ELEMENT))
if 'DRMAA_LIBRARY_PATH' in process_env:
print('DRMAA_LIBRARY_PATH: {}'.format(
process_env['DRMAA_LIBRARY_PATH']))
else:
print(
'DRMAA_LIBRARY_PATH: <NOT_SET>\n\t* Not required unless --distribute is set'
)
# Print snake command
print('Snake command:')
for command_element in snake_command:
if command_element is not None:
print('\t* {}'.format(command_element))
else:
print('\t- None')
# Flush output
sys.stdout.flush()
# Make log directory for distributed jobs
if args.distribute:
os.makedirs('log', exist_ok=True)
# Run snake target
smrtsvrunner.run_snake_target(snakefile, args, process_env, smrtsv_dir,
snake_command)