def main():
"""
Map samples with the selected method(s)
"""
# Load parameters and optional command line input to overwrite config.yaml
params, args = ll.load_params_and_input("map")
mappers = {"bwa": bwa, "bowtie": bowtie}
os.makedirs(params["paths"]["map"], exist_ok=True)
# Returns a list of [path, sampleID] for each sample to assemble
# From command line input (-s or -d or config file)
if args.sample:
samples_to_assemble = [[args.sample, args.sample.rsplit('/', 1)[1]]]
else:
samples_to_assemble = ll.paths_and_samples(
params["paths"]["reads_to_map"])
# Map all samples with all active mapping methods from config file:
for method in params["methods"]["map"]:
for sample in samples_to_map:
outdir = ll.make_dir("{0}{1}".format(params["paths"]["map"],
sample[1]))
job_script = "{0}/{1}.{2}.sh".format(
outdir.rsplit('/', 1)[0], sample[1], method)
# Write job script:
with open(job_script, "w") as file_out:
msub_lines = ll.fill_header(job_script, params)
msub_lines += mappers[method](params, sample, outdir)
for line in msub_lines:
file_out.write(line)
if params["sub"] in (
True, "True", "T", "true",
1): #Move str->bool conversion to ll.parse_argument
ll.submit_job(job_script)
def main():
# Parse arguments
params, args = ll.load_params_and_input(None)
# Map method argument to cleanup function
run_cleanup = {"megahit" : megahit, "idba" : idba}
# Get all samples to cleanup:
cleanup_folders = ll.paths_and_samples(args.directory)
# Run cleanup scripts:
for sample in cleanup_folders:
run_cleanup[args.methods](sample[0].rstrip('/'))
def main():
"""Anotate sample contigs with prokka using default settings + metagenome option"""
params, args = ll.load_params_and_input(None)
os.makedirs(params["paths"]["annotations"], exist_ok=True)
# Annotate the same assemblies we mapped to earlier in the pipeline:
samples_to_annotate = ll.paths_and_samples(params["paths"]["reads_to_map"])
for sample in samples_to_annotate:
outdir = ll.make_dir("{0}{1}".format(params["paths"]["annotations"], sample[1]))
job_script = "{0}/{1}.prokka.sh".format(outdir.rsplit('/',1)[0], sample[1])
# Write job script:
with open(job_script, "w") as file_out:
msub_lines = ll.fill_header(job_script, params)
msub_lines += prokka(params, sample, outdir)
for line in msub_lines:
file_out.write(line)
if params["sub"] in (True, "True", "T", "true", 1): #TODO: Move str->bool conversion to ll.parse_argument
ll.submit_job(job_script)
def main():
"""
Map samples with the selected method(s)
"""
params, args = ll.load_params_and_input("map")
os.makedirs(params["path_out"]["co_assembly_map"], exist_ok=True)
samples_to_map = ll.paths_and_samples(params["paths"]["reads_to_map"])
for sample in samples_to_map:
outdir = ll.make_dir("{0}{1}".format(params["path_out"]["co_assembly_map"], sample[1]))
job_script = "{0}/{1}.{2}.sh".format(outdir.rsplit('/',1)[0], sample[1], "BWA_co_assembly")
# Write job script:
with open(job_script, "w") as file_out:
msub_lines = ll.fill_header(job_script, params)
msub_lines += BWA_co(params, sample, outdir, params["paths"]["co_index"])
for line in msub_lines:
file_out.write(line)
if params["sub"] in (True, "True", "T", "true", 1): #Move str->bool conversion to ll.parse_argument
ll.submit_job(job_script)
def main():
""" Make directories, parse input
Create a job submission script,
Write standard header
Write checkM lineage workflow commands
Submit job if sub parameter is True
"""
params, args = ll.load_params_and_input(None)
os.makedirs(params["paths"]["checkM"], exist_ok=True)
os.makedirs(params["paths"]["jobs"], exist_ok=True)
now = datetime.datetime.strftime(datetime.datetime.now(), '%d_%m_%Y-%H-%M')
job_script = "{0}/checkM_job_{1}.sh".format(params["paths"]["jobs"], now)
with open(job_script, "w") as file_out:
msub_lines = ll.fill_header(job_script, params)
msub_lines += lineage_wf_wrapper(args, params)
for line in msub_lines:
file_out.write(line)
if params["sub"] in (True, "True", "T", "true", 1):
ll.submit_job(job_script)
def main():
params, args = ll.load_params_and_input("assembly")
contig_names = {"idba" : "contig.fa",
"megahit" : "final.contigs.fa"}
# If -sample is passed, just check assembly stats for that file,
# otherwise expand the directory based on config or -directory
if args.sample is None:
quast_list = []
for method in params["methods"]["assembly"]:
samples = ll.paths_and_samples("{0}{1}/".format(params["paths"]["assembly"], method))
full_paths = ["{}/{}".format(x[0], contig_names[method]) for x in samples]
quast_list.extend(full_paths)
files_to_quast = " ".join(quast_list)
else:
file_to_quast = args.sample
if not os.path.exists(params["paths"]["quast_out"]):
os.makedirs(params["paths"]["quast_out"])
timestamp = datetime.datetime.strftime(datetime.datetime.now(), '%Y-%m-%d')
job_script = "{0}{1}.quast.sh".format(params["paths"]["quast_out"], timestamp)
#Actually it's easier to just run on a head node.
header_info = ll.fill_header(job_script, params)
# Could add support/option for metaquast later, I mostly found it useless.
quast_lines = ["{0} {1} -o {2} --threads {3} -L".format(
params["binaries"]["quast"]
files_to_quast,
params["paths"]["quast_out"],
params["num_threads"])]
# Write job script
cmd_lines = header_info + quast_lines
with open(job_script, 'w') as file_out:
for line in cmd_lines:
file_out.write(line)
if params["sub"] is True:
ll.submit_job(job_script)