def fast_denoiser(sff_fps, fasta_fp, tmp_outdir, num_cpus, primer, verbose=True, titanium=False):
"""wrapper function calling methods from the Denoiser package."""
if num_cpus > 1:
denoise_seqs(
sff_fps,
fasta_fp,
tmp_outdir,
primer=primer,
cluster=True,
num_cpus=num_cpus,
verbose=verbose,
titanium=titanium,
)
else:
denoise_seqs(sff_fps, fasta_fp, tmp_outdir, primer=primer, verbose=verbose, titanium=titanium)
# read centroids and singletons
centroids = parse_fasta(open(tmp_outdir + "/centroids.fasta"))
singletons = parse_fasta(open(tmp_outdir + "/singletons.fasta"))
seqs = chain(centroids, singletons)
# read mapping
mapping = {}
cluster_mapping = open(tmp_outdir + "/denoiser_mapping.txt")
for i, cluster in enumerate(cluster_mapping):
cluster, members = cluster.split(":")
members = members.split()
clust = [cluster]
clust.extend(members)
mapping[i] = clust
return seqs, mapping
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 fast_denoiser(
sff_fps, fasta_fp, tmp_outdir, num_cpus, primer, verbose=True,
titanium=False):
"""wrapper function calling methods from the Denoiser package."""
if num_cpus > 1:
denoise_seqs(sff_fps, fasta_fp, tmp_outdir,
primer=primer, cluster=True, num_cpus=num_cpus,
verbose=verbose, titanium=titanium)
else:
denoise_seqs(sff_fps, fasta_fp, tmp_outdir, primer=primer,
verbose=verbose, titanium=titanium)
# read centroids and singletons
centroids = MinimalFastaParser(open(tmp_outdir + "/centroids.fasta"))
singletons = MinimalFastaParser(open(tmp_outdir + "/singletons.fasta"))
seqs = chain(centroids, singletons)
# read mapping
mapping = {}
cluster_mapping = open(tmp_outdir + "/denoiser_mapping.txt")
for i, cluster in enumerate(cluster_mapping):
cluster, members = cluster.split(':')
members = members.split()
clust = [cluster]
clust.extend(members)
mapping[i] = clust
return seqs, mapping
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)
