def main(commandline_args=None):
parser, opts, args = parse_command_line_parameters(**script_info)
if not opts.sff_fp:
parser.error('Required option flowgram file path (-i) not specified')
elif not files_exist(opts.sff_fp):
parser.error('Flowgram file path does not exist:\n %s \n Pass a valid one via -i.'
% opts.sff_fp)
if(opts.checkpoint_fp):
bp_fp = opts.checkpoint_fp
if not exists(bp_fp):
parser.error('Specified checkpoint file does not exist: %s' % bp_fp)
#peek into sff.txt files to make sure they are parseable
#cat_sff_fles is lazy and only reads header
flowgrams, header = cat_sff_files(map(open, opts.sff_fp.split(',')))
if(opts.split and opts.preprocess_fp):
parser.error('Options --split and --preprocess_fp are exclusive')
if(opts.preprocess_fp):
pp_fp = opts.preprocess_fp
if not exists(opts.preprocess_fp):
parser.error('Specified preprocess directory does not exist: %s' % opts.preprocess_fp)
if not files_exist('%s/prefix_mapping.txt,%s/prefix_dereplicated.fasta' %(pp_fp, pp_fp)):
parser.error('Specified preprocess directory does not contain expected files: ' +\
'prefix_mapping.txt and prefix_dereplicated.fasta')
if opts.titanium:
opts.error_profile = DENOISER_DATA_DIR+'Titanium_error_profile.dat'
opts.low_cutoff = 4
opts.high_cutoff = 5
if not exists(opts.error_profile):
parser.error('Specified error profile %s does not exist' % opts.error_profile)
if opts.output_dir:
#make sure it always ends on /
tmpoutdir=opts.output_dir+"/"
else:
#make random dir in current dir
tmpoutdir = get_tmp_filename(tmp_dir="", prefix="denoiser_", suffix="/")
create_dir(tmpoutdir, not opts.force)
log_fp = 'denoiser.log'
if opts.split:
denoise_per_sample(opts.sff_fp, opts.fasta_fp, tmpoutdir, opts.cluster,
opts.num_cpus, opts.squeeze, opts.percent_id, opts.bail,
opts.primer, opts.low_cutoff, opts.high_cutoff, log_fp,
opts.low_memory, opts.verbose, opts.error_profile, opts.max_num_iter,
opts.titanium)
else:
denoise_seqs(opts.sff_fp, opts.fasta_fp, tmpoutdir, opts.preprocess_fp, opts.cluster,
opts.num_cpus, opts.squeeze, opts.percent_id, opts.bail, opts.primer,
opts.low_cutoff, opts.high_cutoff, log_fp, opts.low_memory,
opts.verbose, opts.error_profile, opts.max_num_iter, opts.titanium,
opts.checkpoint_fp)
def main(commandline_args=None):
parser, opts, args = parse_command_line_parameters(**script_info)
if not opts.sff_fp:
parser.error('Required option flowgram file path (-i) not specified')
elif not files_exist(opts.sff_fp):
parser.error('Flowgram file path does not exist:\n %s \n Pass a valid one via -i.'
% opts.sff_fp)
if(opts.checkpoint_fp):
bp_fp = opts.checkpoint_fp
if not exists(bp_fp):
parser.error('Specified checkpoint file does not exist: %s' % bp_fp)
#peek into sff.txt files to make sure they are parseable
#cat_sff_fles is lazy and only reads header
flowgrams, header = cat_sff_files(map(open, opts.sff_fp.split(',')))
if(opts.split and opts.preprocess_fp):
parser.error('Options --split and --preprocess_fp are exclusive')
if(opts.preprocess_fp):
pp_fp = opts.preprocess_fp
if not exists(opts.preprocess_fp):
parser.error('Specified preprocess directory does not exist: %s' % opts.preprocess_fp)
if not files_exist('%s/prefix_mapping.txt,%s/prefix_dereplicated.fasta' %(pp_fp, pp_fp)):
parser.error('Specified preprocess directory does not contain expected files: ' +\
'prefix_mapping.txt and prefix_dereplicated.fasta')
if opts.titanium:
opts.error_profile = DENOISER_DATA_DIR+'Titanium_error_profile.dat'
opts.low_cutoff = 4
opts.high_cutoff = 5
if not exists(opts.error_profile):
parser.error('Specified error profile %s does not exist' % opts.error_profile)
if opts.output_dir:
#make sure it always ends on /
tmpoutdir=opts.output_dir+"/"
else:
#make random dir in current dir
tmpoutdir = get_tmp_filename(tmp_dir="", prefix="denoiser_", suffix="/")
create_dir(tmpoutdir, not opts.force)
log_fp = 'denoiser.log'
if opts.split:
denoise_per_sample(opts.sff_fp, opts.fasta_fp, tmpoutdir, opts.cluster,
opts.num_cpus, opts.squeeze, opts.percent_id, opts.bail,
opts.primer, opts.low_cutoff, opts.high_cutoff, log_fp,
opts.low_memory, opts.verbose, opts.error_profile, opts.max_num_iter,
opts.titanium)
else:
denoise_seqs(opts.sff_fp, opts.fasta_fp, tmpoutdir, opts.preprocess_fp, opts.cluster,
opts.num_cpus, opts.squeeze, opts.percent_id, opts.bail, opts.primer,
opts.low_cutoff, opts.high_cutoff, log_fp, opts.low_memory,
opts.verbose, opts.error_profile, opts.max_num_iter, opts.titanium,
opts.checkpoint_fp)
def main(commandline_args=None):
parser, opts, args = parse_command_line_parameters(**script_info)
if not opts.sff_fp:
parser.error('Required option flowgram file path (-i) not specified')
elif not files_exist(opts.sff_fp):
parser.error(
'Flowgram file path does not exist:\n %s \n Pass a valid one via -i.'
% opts.sff_fp)
#make tmp and output dir
tmp_dir = get_tmp_filename(tmp_dir=opts.output_dir + "/", suffix="/")
try:
makedirs(tmp_dir)
except OSError:
exit("Creating temporary directory failed")
if (not exists(opts.output_dir)):
try:
makedirs(opts.output_dir)
except OSError:
exit("Creating output directory failed")
#open logger
log_fh = None
if opts.verbose:
#append to the log file of the master process
log_fh = open(opts.output_dir + "/" + opts.log_fp, "a", 0)
log_fh.write("SFF file: %s\n" % opts.sff_fp)
log_fh.write("Fasta file: %s\n" % opts.fasta_fp)
log_fh.write("Output dir: %s\n" % opts.output_dir)
log_fh.write("Squeeze Seqs: %s\n" % opts.squeeze)
log_fh.write("Primer sequence: %s\n" % opts.primer)
(deprefixed_sff_fp, l, mapping, seqs) = \
preprocess(opts.sff_fp, log_fh, fasta_fp=opts.fasta_fp,
out_fp=tmp_dir,
verbose=opts.verbose, squeeze=opts.squeeze,
primer=opts.primer)
# explicitly close log file, as this file can be shared with the master
# Closing it here assures that all preprocess writes happen before the
# master writes
if log_fh:
log_fh.close()
#move files to output dir
rename(tmp_dir + "/prefix_dereplicated.sff.txt",
opts.output_dir + "/prefix_dereplicated.sff.txt")
rename(tmp_dir + "/prefix_dereplicated.fasta",
opts.output_dir + "/prefix_dereplicated.fasta")
rename(tmp_dir + "/prefix_mapping.txt",
opts.output_dir + "/prefix_mapping.txt")
rmdir(tmp_dir)
def main(commandline_args=None):
parser, opts, args = parse_command_line_parameters(**script_info)
if not opts.sff_fp:
parser.error('Required option flowgram file path (-i) not specified')
elif not files_exist(opts.sff_fp):
parser.error('Flowgram file path does not exist:\n %s \n Pass a valid one via -i.'
% opts.sff_fp)
#make tmp and output dir
tmp_dir = get_tmp_filename(tmp_dir = opts.output_dir+"/", suffix="/")
try:
makedirs(tmp_dir)
except OSError:
exit("Creating temporary directory failed")
if(not exists(opts.output_dir)):
try:
makedirs(opts.output_dir)
except OSError:
exit("Creating output directory failed")
#open logger
log_fh=None
if opts.verbose:
#append to the log file of the master process
log_fh = open(opts.output_dir+"/"+opts.log_fp, "a", 0)
log_fh.write("SFF file: %s\n" % opts.sff_fp)
log_fh.write("Fasta file: %s\n" % opts.fasta_fp)
log_fh.write("Output dir: %s\n" % opts.output_dir)
log_fh.write("Squeeze Seqs: %s\n" % opts.squeeze)
log_fh.write("Primer sequence: %s\n" % opts.primer)
(deprefixed_sff_fp, l, mapping, seqs) = \
preprocess(opts.sff_fp, log_fh, fasta_fp=opts.fasta_fp,
out_fp=tmp_dir,
verbose=opts.verbose, squeeze=opts.squeeze,
primer=opts.primer)
# explicitly close log file, as this file can be shared with the master
# Closing it here assures that all preprocess writes happen before the
# master writes
if log_fh:
log_fh.close()
#move files to output dir
rename(tmp_dir+"/prefix_dereplicated.sff.txt",
opts.output_dir+"/prefix_dereplicated.sff.txt")
rename(tmp_dir+"/prefix_dereplicated.fasta",
opts.output_dir+"/prefix_dereplicated.fasta")
rename(tmp_dir+"/prefix_mapping.txt", opts.output_dir+"/prefix_mapping.txt")
rmdir(tmp_dir)
