def analyse(ctx, sampleinfo_file, input, config, dry, email, skip_update,
force_update, untrimmed, uncareful):
"""Sequence analysis, typing and resistance identification"""
# Run section
pool = []
trimmed = not untrimmed
careful = not uncareful
set_cli_config(config)
ctx.obj["config"]["regex"]["mail_recipient"] = email
ctx.obj["config"]["dry"] = dry
if not os.path.isdir(input):
click.echo(
"ERROR - Sequence data folder {} does not exist.".format(input))
ctx.abort()
for subfolder in os.listdir(input):
if os.path.isdir("{}/{}".format(input, subfolder)):
pool.append(subfolder)
run_settings = {
"input": input,
"dry": dry,
"email": email,
"skip_update": skip_update,
"trimmed": not untrimmed,
"careful": not uncareful,
"pool": pool,
}
# Samples section
sampleinfo = review_sampleinfo(sampleinfo_file)
run_creator = Job_Creator(
config=ctx.obj["config"],
log=ctx.obj["log"],
sampleinfo=sampleinfo,
run_settings=run_settings,
)
ext_refs = Referencer(config=ctx.obj["config"],
log=ctx.obj["log"],
sampleinfo=sampleinfo,
force=force_update)
click.echo("INFO - Checking versions of references..")
try:
if not skip_update:
ext_refs.identify_new(project=True)
ext_refs.update_refs()
click.echo("INFO - Version check done. Creating sbatch jobs")
else:
click.echo("INFO - Skipping version check.")
except Exception as e:
click.echo("{}".format(e))
if len(sampleinfo) > 1:
run_creator.project_job()
elif len(sampleinfo) == 1:
run_creator.project_job(single_sample=True)
else:
ctx.abort()
done()
def autobatch(ctx, dry, skip_update, email):
"""Analyses all currently unanalysed projects in the seqdata folder"""
# Trace exists by some samples having pubMLST_ST filled in. Make trace function later
ctx.obj["config"]["regex"]["mail_recipient"] = email
ext_refs = Referencer(config=ctx.obj["config"], log=ctx.obj["log"])
if not skip_update:
ext_refs.identify_new(project=True)
ext_refs.update_refs()
process = subprocess.Popen(
'squeue --format="%50j" -h -r'.split(), stdout=subprocess.PIPE
)
run_jobs, error = process.communicate()
run_jobs = run_jobs.splitlines()
run_jobs = [
re.search('"(.+)"', str(jobname)).group(1).replace(" ", "")
for jobname in run_jobs
]
old_jobs = os.listdir(ctx.obj["config"]["folders"]["results"])
for foldah in os.listdir(ctx.obj["config"]["folders"]["seqdata"]):
# Project name not found in slurm list
if len([job for job in run_jobs if foldah in job]) == 0:
# Project name not found in results directories
if len([job for job in old_jobs if foldah in job]) == 0:
if dry:
click.echo(
"DRY - microSALT analyse {} --skip_update".format(foldah)
)
else:
process = subprocess.Popen(
"microSALT analyse project {} --skip_update".format(
foldah
).split(),
stdout=subprocess.PIPE,
)
output, error = process.communicate()
elif dry:
click.echo(
"INFO - Skipping {} due to existing analysis in results folder".format(
foldah
)
)
elif dry:
click.echo("INFO - Skipping {} due to concurrent SLURM run".format(foldah))
done()
def init_references(testdata):
ref_obj = Referencer(config=preset_config, log=logger, sampleinfo=testdata)
ref_obj.identify_new(testdata[0].get('CG_ID_project'), project=True)
ref_obj.update_refs()
def finish(ctx, sampleinfo_file, input, track, config, dry, email, skip_update,
report, output):
"""Sequence analysis, typing and resistance identification"""
# Run section
pool = []
set_cli_config(config)
ctx.obj["config"]["regex"]["mail_recipient"] = email
ctx.obj["config"]["dry"] = dry
if not os.path.isdir(input):
click.echo(
"ERROR - Sequence data folder {} does not exist.".format(input))
ctx.abort()
if output == "":
output = input
for subfolder in os.listdir(input):
if os.path.isdir("{}/{}".format(input, subfolder)):
pool.append(subfolder)
run_settings = {
"input": input,
"track": track,
"dry": dry,
"email": email,
"skip_update": skip_update,
}
# Samples section
sampleinfo = review_sampleinfo(sampleinfo_file)
ext_refs = Referencer(config=ctx.obj["config"],
log=ctx.obj["log"],
sampleinfo=sampleinfo)
click.echo("INFO - Checking versions of references..")
try:
if not skip_update:
ext_refs.identify_new(project=True)
ext_refs.update_refs()
click.echo("INFO - Version check done. Creating sbatch jobs")
else:
click.echo("INFO - Skipping version check.")
except Exception as e:
click.echo("{}".format(e))
res_scraper = Scraper(config=ctx.obj["config"],
log=ctx.obj["log"],
sampleinfo=sampleinfo,
input=input)
if isinstance(sampleinfo, list) and len(sampleinfo) > 1:
res_scraper.scrape_project()
# for subfolder in pool:
# res_scraper.scrape_sample()
else:
res_scraper.scrape_sample()
codemonkey = Reporter(
config=ctx.obj["config"],
log=ctx.obj["log"],
sampleinfo=sampleinfo,
output=output,
collection=True,
)
codemonkey.report(report)
done()