def setup(args):
template, template_txt = name_to_config(args.template)
base_item = template["details"][0]
project_name, metadata, global_vars = _pname_and_metadata(args.metadata)
remotes = _retrieve_remote([args.template, args.metadata])
inputs = args.input_files + remotes.get("inputs", [])
items = [
_add_metadata(item, metadata, remotes)
for item in _prep_items_from_base(base_item, inputs)
]
out_dir = os.path.join(os.getcwd(), project_name)
work_dir = utils.safe_makedir(os.path.join(out_dir, "work"))
if len(items) == 0:
out_config_file = _write_template_config(template_txt, project_name,
out_dir)
print "Template configuration file created at: %s" % out_config_file
print "Edit to finalize custom options, then prepare full sample config with:"
print " bcbio_nextgen.py -w template %s %s sample1.bam sample2.fq" % \
(out_config_file, project_name)
else:
out_config_file = _write_config_file(items, global_vars, template,
project_name, out_dir)
print "Configuration file created at: %s" % out_config_file
print "Edit to finalize and run with:"
print " cd %s" % work_dir
print " bcbio_nextgen.py ../config/%s" % os.path.basename(
out_config_file)
if remotes.get("base"):
remote_path = os.path.join(remotes["base"],
os.path.basename(out_config_file))
bucket, key = utils.s3_bucket_key(remote_path)
s3.upload_file_boto(out_config_file, bucket, key)
print "Also uploaded to AWS S3 in %s" % remotes["base"]
print "Run directly with bcbio_vm.py run %s" % remote_path
def setup(args):
template, template_txt = name_to_config(args.template)
base_item = template["details"][0]
project_name, metadata, global_vars = _pname_and_metadata(args.metadata)
remotes = _retrieve_remote([args.template, args.metadata])
inputs = args.input_files + remotes.get("inputs", [])
items = [_add_metadata(item, metadata, remotes)
for item in _prep_items_from_base(base_item, inputs)]
out_dir = os.path.join(os.getcwd(), project_name)
work_dir = utils.safe_makedir(os.path.join(out_dir, "work"))
if len(items) == 0:
out_config_file = _write_template_config(template_txt, project_name, out_dir)
print "Template configuration file created at: %s" % out_config_file
print "Edit to finalize custom options, then prepare full sample config with:"
print " bcbio_nextgen.py -w template %s %s sample1.bam sample2.fq" % \
(out_config_file, project_name)
else:
out_config_file = _write_config_file(items, global_vars, template, project_name, out_dir)
print "Configuration file created at: %s" % out_config_file
print "Edit to finalize and run with:"
print " cd %s" % work_dir
print " bcbio_nextgen.py ../config/%s" % os.path.basename(out_config_file)
if remotes.get("base"):
remote_path = os.path.join(remotes["base"], os.path.basename(out_config_file))
bucket, key = utils.s3_bucket_key(remote_path)
s3.upload_file_boto(out_config_file, bucket, key)
print "Also uploaded to AWS S3 in %s" % remotes["base"]
print "Run directly with bcbio_vm.py run %s" % remote_path
def _retrieve_remote(fnames):
"""Retrieve remote inputs found in the same bucket as the template or metadata files.
"""
for fname in fnames:
if fname.startswith(utils.SUPPORTED_REMOTES):
bucket_name, key_name = utils.s3_bucket_key(fname)
folder = os.path.dirname(key_name)
conn = boto.connect_s3()
bucket = conn.get_bucket(bucket_name)
inputs = []
regions = []
for key in bucket.get_all_keys(prefix=folder):
if key.name.endswith(tuple(KNOWN_EXTS.keys())):
inputs.append("s3://%s/%s" % (bucket_name, key.name))
elif key.name.endswith((".bed", ".bed.gz")):
regions.append("s3://%s/%s" % (bucket_name, key.name))
remote_base = "s3://{bucket_name}/{folder}".format(**locals())
return {"base": remote_base,
"inputs": inputs,
"region": regions[0] if len(regions) == 1 else None}
return {}
def load_s3(sample_config):
"""Move a sample configuration locally, providing remote upload.
"""
with utils.s3_handle(sample_config) as in_handle:
config = yaml.load(in_handle)
bucket, key = utils.s3_bucket_key(sample_config)
config["upload"] = {"method": "s3",
"dir": os.path.join(os.pardir, "final"),
"bucket": bucket,
"folder": os.path.join(os.path.dirname(key), "final")}
if not os.access(os.pardir, os.W_OK | os.X_OK):
raise IOError("Cannot write to the parent directory of work directory %s\n"
"bcbio wants to store prepared uploaded files to %s\n"
"We recommend structuring your project in a project specific directory structure\n"
"with a specific work directory (mkdir -p your-project/work && cd your-project/work)."
% (os.getcwd(), os.path.join(os.pardir, "final")))
config = _add_jar_resources(config, bucket, key)
out_file = os.path.join(utils.safe_makedir(os.path.join(os.getcwd(), "config")),
os.path.basename(key))
with open(out_file, "w") as out_handle:
yaml.dump(config, out_handle, default_flow_style=False, allow_unicode=False)
return out_file
def _retrieve_remote(fnames):
"""Retrieve remote inputs found in the same bucket as the template or metadata files.
"""
for fname in fnames:
if fname.startswith(utils.SUPPORTED_REMOTES):
bucket_name, key_name = utils.s3_bucket_key(fname)
folder = os.path.dirname(key_name)
conn = boto.connect_s3()
bucket = conn.get_bucket(bucket_name)
inputs = []
regions = []
for key in bucket.get_all_keys(prefix=folder):
if key.name.endswith(tuple(KNOWN_EXTS.keys())):
inputs.append("s3://%s/%s" % (bucket_name, key.name))
elif key.name.endswith((".bed", ".bed.gz")):
regions.append("s3://%s/%s" % (bucket_name, key.name))
remote_base = "s3://{bucket_name}/{folder}".format(**locals())
return {
"base": remote_base,
"inputs": inputs,
"region": regions[0] if len(regions) == 1 else None
}
return {}