def test_add_runner_option_with_argparse(self):
"""add_runner_option enables '--runner' with ArgumentParser
"""
p = ArgumentParser()
add_runner_option(p)
args = p.parse_args(['--runner', 'SimpleJobRunner'])
self.assertEqual(args.runner, 'SimpleJobRunner')
def test_add_runner_option(self):
"""add_runner_option enables '--runner'
"""
# Skip the test if optparse not available
if not OPTPARSE_AVAILABLE:
raise unittest.SkipTest("'optparse' not available")
p = OptionParser()
add_runner_option(p)
options, args = p.parse_args(['--runner', 'SimpleJobRunner'])
self.assertEqual(options.runner, 'SimpleJobRunner')
def add_analyse_barcodes_command(cmdparser):
"""Create a parser for the 'analyse_barcodes' command
"""
p = cmdparser.add_command('analyse_barcodes',help="Analyse index (barcode) sequences",
usage="%prog analyse_barcodes [OPTIONS] [ANALYSIS_DIR]",
version="%prog "+__version__,
description="Analyse barcode sequences for fastq files "
"in specified lanes in ANALYSIS_DIR, and report the most "
"common barcodes found across all reads from each lane.")
p.add_option('--unaligned-dir',action='store',
dest='unaligned_dir',default='bcl2fastq',
help="explicitly set the (sub)directory with bcl-to-fastq outputs")
p.add_option('--lanes',action='store',
dest='lanes',default=None,
help="specify which lanes to analyse barcodes for (default is to do "
"analysis for all lanes).")
p.add_option('--mismatches',action='store',dest='mismatches',
default=0,type='int',
help="maximum number of mismatches to use when grouping "
"similar barcodes (default is 0, i.e. no grouping)")
p.add_option('--cutoff',action='store',dest='cutoff',
default=0.001,type='float',
help="exclude barcodes with a smaller fraction of "
"associated reads than CUTOFF, e.g. '0.01' excludes "
"barcodes with < 1% of reads (default is 0.01%)")
p.add_option('--sample-sheet',action="store",
dest="sample_sheet",default=None,
help="use an alternative sample sheet to the default "
"'custom_SampleSheet.csv' created on setup.")
p.add_option('--barcode-analysis-dir',action="store",
dest="barcode_analysis_dir",default=None,
help="specify subdirectory where barcode analysis will "
"be performed and outputs will be written")
add_runner_option(p)
add_debug_option(p)
# Deprecated options
deprecated = optparse.OptionGroup(p,'Deprecated/defunct options')
deprecated.add_option('--nprocessors',action='store',
dest='nprocessors',default=None,type='int',
help="does nothing; kept for backwards "
"compatibility only")
deprecated.add_option('--truncate',action='store',
dest='length',default=None,type='int',
help="does nothing; kept for backwards "
"compatibility only")
p.add_option_group(deprecated)
def add_update_fastq_stats_command(cmdparser):
"""Create a parser for the 'update_fastq_stats' command
"""
p = cmdparser.add_command('update_fastq_stats',help="(Re)generate Fastq statistics",
usage="%prog update_fastq_stats [OPTIONS] [ANALYSIS_DIR]",
description="(Re)generate statistics for fastq "
"files produced from 'make_fastqs'.")
p.add_option('--unaligned-dir',action='store',
dest='unaligned_dir',default='bcl2fastq',
help="explicitly set the (sub)directory with bcl-to-fastq outputs")
p.add_option('--stats-file',action='store',
dest='stats_file',default=None,
help="specify output file for fastq statistics")
p.add_option('--per-lane-stats-file',action='store',
dest='per_lane_stats_file',default=None,
help="specify output file for per-lane statistics")
add_nprocessors_option(p,__settings.fastq_stats.nprocessors)
add_runner_option(p)
add_debug_option(p)
def add_run_qc_command(cmdparser):
"""Create a parser for the 'run_qc' command
"""
p = cmdparser.add_command('run_qc',help="Run QC procedures",
usage="%prog run_qc [OPTIONS] [ANALYSIS_DIR]",
description="Run QC procedures for sequencing projects in "
"ANALYSIS_DIR.")
max_concurrent_jobs = __settings.general.max_concurrent_jobs
fastq_screen_subset = 1000000
p.add_option('--projects',action='store',
dest='project_pattern',default=None,
help="simple wildcard-based pattern specifying a subset of projects "
"and samples to run the QC on. PROJECT_PATTERN should be of the form "
"'pname[/sname]', where 'pname' specifies a project (or set of "
"projects) and 'sname' optionally specifies a sample (or set of "
"samples).")
p.add_option('--fastq_screen_subset',action='store',dest='subset',
type='int',default=fastq_screen_subset,
help="specify size of subset of total reads to use for "
"fastq_screen (i.e. --subset option); (default %d, set to "
"0 to use all reads)" % fastq_screen_subset)
p.add_option('--ungzip-fastqs',action='store_true',dest='ungzip_fastqs',
help="create decompressed copies of fastq.gz files")
p.add_option('--max-jobs',action='store',
dest='max_jobs',default=max_concurrent_jobs,type='int',
help="explicitly specify maximum number of concurrent QC jobs to run "
"(default %s, change in settings file)" % max_concurrent_jobs)
add_runner_option(p)
add_modulefiles_option(p)
add_debug_option(p)
# Deprecated options
deprecated = optparse.OptionGroup(p,'Deprecated/defunct options')
deprecated.add_option('--no-ungzip-fastqs',action='store_true',dest='no_ungzip_fastqs',
help="don't create uncompressed copies of fastq.gz files "
"(does nothing; this is now the default, use --ungzip-fastqs "
"to turn on decompression)")
p.add_option_group(deprecated)
def add_make_fastqs_command(cmdparser):
"""Create a parser for the 'make_fastqs' command
"""
p = cmdparser.add_command('make_fastqs',help="Run Fastq generation",
usage="%prog make_fastqs [OPTIONS] [ANALYSIS_DIR]",
description="Generate fastq files from raw bcl files "
"produced by Illumina sequencer within ANALYSIS_DIR.")
# General options
add_no_save_option(p)
add_modulefiles_option(p)
add_debug_option(p)
# Primary data management
primary_data = optparse.OptionGroup(p,'Primary data management')
primary_data.add_option('--only-fetch-primary-data',action='store_true',
dest='only_fetch_primary_data',default=False,
help="only fetch the primary data, don't perform any other "
"operations")
primary_data.add_option('--skip-rsync',action='store_true',
dest='skip_rsync',default=False,
help="don't rsync the primary data at the beginning of processing")
primary_data.add_option('--remove-primary-data',action='store_true',
dest='remove_primary_data',default=False,
help="Delete the primary data at the end of processing (default "
"is to keep data)")
p.add_option_group(primary_data)
# Options to control bcl2fastq
bcl_to_fastq = optparse.OptionGroup(p,'Bcl-to-fastq options')
bcl_to_fastq.add_option('--skip-bcl2fastq',action='store_true',
dest='skip_bcl2fastq',default=False,
help="don't run the Fastq generation step")
bcl_to_fastq.add_option('--output-dir',action='store',
dest='unaligned_dir',default=None,
help="explicitly set the output (sub)directory for bcl-to-fastq "
"conversion (overrides default)")
bcl_to_fastq.add_option('--use-bases-mask',action="store",
dest="bases_mask",default=None,
help="explicitly set the bases-mask string to indicate how each "
"cycle should be used in the bcl-to-fastq conversion (overrides "
"default)")
bcl_to_fastq.add_option('--sample-sheet',action="store",
dest="sample_sheet",default=None,
help="use an alternative sample sheet to the default "
"'custom_SampleSheet.csv' created on setup.")
bcl_to_fastq.add_option('--ignore-missing-bcl',action='store_true',
dest='ignore_missing_bcl',default=False,
help="use the --ignore-missing-bcl option for bcl2fastq (treat "
"missing bcl files as no call)")
bcl_to_fastq.add_option('--ignore-missing-stats',action='store_true',
dest='ignore_missing_stats',default=False,
help="use the --ignore-missing-stats option for bcl2fastq (fill "
"in with zeroes when *.stats files are missing)")
bcl_to_fastq.add_option('--require-bcl2fastq-version',action='store',
dest='bcl2fastq_version',default=None,
help="explicitly specify version of bcl2fastq "
"software to use (e.g. '1.8.4' or '>=2.0').")
# Use lane splitting
# Determine defaults to report to user
no_lane_splitting_platforms = []
use_lane_splitting_platforms = []
for platform in __settings.platform:
if __settings.platform[platform].no_lane_splitting is not None:
if __settings.platform[platform].no_lane_splitting:
no_lane_splitting_platforms.append(platform)
else:
use_lane_splitting_platforms.append(platform)
if __settings.bcl2fastq.no_lane_splitting:
if use_lane_splitting_platforms:
default_no_lane_splitting = \
"Used by default for all platforms except %s" % \
', '.join(use_lane_splitting_platforms)
default_use_lane_splitting = "Used by default for %s" % \
', '.join(use_lane_splitting_platforms)
else:
default_no_lane_splitting = "Default for all platforms"
default_use_lane_splitting = ""
else:
if no_lane_splitting_platforms:
default_use_lane_splitting = \
"Used by default for all platforms except %s" % \
', '.join(no_lane_splitting_platforms)
default_no_lane_splitting = "Used by default for %s" % \
', '.join(no_lane_splitting_platforms)
else:
default_no_lane_splitting = ""
default_use_lane_splitting = "Default for all platforms"
if default_use_lane_splitting:
default_use_lane_splitting = ". "+default_use_lane_splitting
if default_no_lane_splitting:
default_no_lane_splitting = ". "+default_no_lane_splitting
bcl_to_fastq.add_option('--no-lane-splitting',action='store_true',
dest='no_lane_splitting',default=False,
help="don't split the output FASTQ files by lane "
"(bcl2fastq v2 only; turn off using "
"--use-lane-splitting)%s" % default_no_lane_splitting)
bcl_to_fastq.add_option('--use-lane-splitting',action='store_true',
dest='use_lane_splitting',default=False,
help="split the output FASTQ files by lane "
"(bcl2fastq v2 only; turn off using "
"--no-lane-splitting)%s" % default_use_lane_splitting)
# Adapter trimming/masking options
bcl_to_fastq.add_option('--minimum-trimmed-read-length',action="store",
dest="minimum_trimmed_read_length",default=35,
help="Minimum read length after adapter "
"trimming. bcl2fastq trims the adapter from "
"the read down to this value; if there is more "
"adapter match below this length then those "
"bases are masked not trimmed (i.e. replaced "
"by N rather than removed) (default: 35)")
bcl_to_fastq.add_option('--mask-short-adapter-reads',action="store",
dest="mask_short_adapter_reads",default=22,
help="minimum length of unmasked bases that "
"a read can be after adapter trimming; reads "
"with fewer ACGT bases will be completely "
"masked with Ns (default: 22)")
# Creation of empty fastqs
bcl_to_fastq.add_option('--create-empty-fastqs',action='store_true',
dest='create_empty_fastqs',default=False,
help="create 'empty' FASTQ files which weren't "
"generated by bcl2fastq because they didn't "
"have any reads assigned at the demultiplexing "
"stage (NB bcl2fastq must have finished without "
"an error for this option to be applied)")
# Number of processors
default_nprocessors = []
for platform in __settings.platform:
if __settings.platform[platform].nprocessors is not None:
default_nprocessors.append("%s: %s" %
(platform,
__settings.platform[platform].nprocessors))
if default_nprocessors:
default_nprocessors.append("other platforms: %s" %
__settings.bcl2fastq.nprocessors)
else:
default_nprocessors.append("%s" % __settings.bcl2fastq.nprocessors)
default_nprocessors = ', '.join(default_nprocessors)
add_nprocessors_option(bcl_to_fastq,None,
default_display=default_nprocessors)
add_runner_option(bcl_to_fastq)
p.add_option_group(bcl_to_fastq)
# Statistics
statistics = optparse.OptionGroup(p,'Statistics generation')
statistics.add_option('--stats-file',action='store',
dest='stats_file',default=None,
help="specify output file for fastq statistics")
statistics.add_option('--per-lane-stats-file',action='store',
dest='per_lane_stats_file',default=None,
help="specify output file for per-lane statistics")
statistics.add_option('--no-stats',action='store_true',
dest='no_stats',default=False,
help="don't generate statistics file; use 'update_fastq_stats' "
"command to (re)generate statistics")
p.add_option_group(statistics)
# Deprecated options
deprecated = optparse.OptionGroup(p,'Deprecated/defunct options')
deprecated.add_option('--keep-primary-data',action='store_true',
dest='keep_primary_data',default=False,
help="don't delete the primary data at the end of processing "
"(does nothing; primary data is kept by default unless "
"--remove-primary-data is specified)")
deprecated.add_option('--generate-stats',action='store_true',
dest='generate_stats',default=False,
help="(re)generate statistics for fastq files (does nothing; "
"statistics are generated by default unless suppressed by "
"--no-stats)")
deprecated.add_option('--report-barcodes',action='store_true',
dest='report_barcodes',default=False,
help="analyse and report barcode indices for all lanes after "
"generating fastq files (deprecated: use the "
"'analyse_barcodes' command instead)")
deprecated.add_option('--barcodes-file',action='store',
dest='barcodes_file',default=None,
help="specify output file for barcode analysis report "
"(deprecated: use the 'analyse_barcodes' command instead)")
p.add_option_group(deprecated)
