def compute_k_gwa_with_loco_and_cavar(chromosomes, maf, token):
"""k-gwa-compute; Loco with covars; lmm defaults to 9!
"""
working_dir = os.path.join(current_app.config.get("TMPDIR"), token)
_dict = jsonfile_to_dict(os.path.join(working_dir, "metadata.json"))
try:
phenofile, snpsfile, covarfile = [
os.path.join(working_dir, _dict.get(x))
for x in ["pheno", "snps", "covar"]
]
genofile = cache_ipfs_file(
ipfs_file=_dict.get("geno"),
cache_dir=current_app.config.get('CACHEDIR'))
if not do_paths_exist([genofile, phenofile, snpsfile]):
raise FileNotFoundError
gemma_kwargs = {"g": genofile, "p": phenofile, "a": snpsfile}
gemma_k_cmd = generate_gemma_cmd(
gemma_cmd=current_app.config.get("GEMMA_"
"WRAPPER_CMD"),
output_dir=current_app.config.get('TMPDIR'),
token=token,
gemma_kwargs=gemma_kwargs,
chromosomes=chromosomes)
gemma_kwargs["c"] = covarfile
gemma_kwargs["maf"] = float(maf)
gemma_kwargs["lmm"] = _dict.get("lmm", 9)
gemma_gwa_cmd = generate_gemma_cmd(
gemma_cmd=current_app.config.get("GEMMA_"
"WRAPPER_CMD"),
output_dir=current_app.config.get('TMPDIR'),
token=token,
gemma_kwargs=gemma_kwargs,
gemma_wrapper_kwargs={
"loco":
("--input "
f"{os.path.join(working_dir, gemma_k_cmd.get('output_file'))}"
)
})
return jsonify(unique_id=queue_cmd(
conn=redis.Redis(),
email=(request.get_json() or {}).get('email'),
job_queue=current_app.config.get("REDIS_JOB_QUEUE"),
cmd=(f"{gemma_k_cmd.get('gemma_cmd')} && "
f"{gemma_gwa_cmd.get('gemma_cmd')}")),
status="queued",
output_file=gemma_gwa_cmd.get("output_file"))
# pylint: disable=W0703
except Exception:
return jsonify(
status=128,
# use better message
message="Metadata file non-existent!")
def compute_k_gwa_with_covars_only(token):
"""Given a genofile, traitfile, snpsfile, and the token, compute the k-values
and return <hash-of-inputs>.json with a UNIQUE-ID of the job. The genofile,
traitfile, and snpsfile are extracted from a metadata.json file. No Loco no
covars; lmm defaults to 9!
"""
working_dir = os.path.join(current_app.config.get("TMPDIR"), token)
_dict = jsonfile_to_dict(os.path.join(working_dir, "metadata.json"))
try:
phenofile, snpsfile, covarfile = [
os.path.join(working_dir, _dict.get(x))
for x in ["pheno", "snps", "covar"]
]
genofile = cache_ipfs_file(
ipfs_file=_dict.get("geno"),
cache_dir=current_app.config.get('CACHEDIR'))
if not do_paths_exist([genofile, phenofile, snpsfile]):
raise FileNotFoundError
gemma_kwargs = {"g": genofile, "p": phenofile, "a": snpsfile}
gemma_k_cmd = generate_gemma_cmd(
gemma_cmd=current_app.config.get("GEMMA_"
"WRAPPER_CMD"),
output_dir=current_app.config.get('TMPDIR'),
token=token,
gemma_kwargs=gemma_kwargs)
gemma_kwargs["c"] = covarfile
gemma_kwargs["lmm"] = _dict.get("lmm", 9)
gemma_gwa_cmd = generate_gemma_cmd(
gemma_cmd=current_app.config.get("GEMMA_"
"WRAPPER_CMD"),
output_dir=current_app.config.get('TMPDIR'),
token=token,
gemma_kwargs=gemma_kwargs,
gemma_wrapper_kwargs={
"input": os.path.join(working_dir,
gemma_k_cmd.get("output_file"))
})
return jsonify(unique_id=queue_cmd(
conn=redis.Redis(),
email=(request.get_json() or {}).get('email'),
job_queue=current_app.config.get("REDIS_JOB_QUEUE"),
cmd=(f"{gemma_k_cmd.get('gemma_cmd')} && "
f"{gemma_gwa_cmd.get('gemma_cmd')}")),
status="queued",
output_file=gemma_gwa_cmd.get("output_file"))
# pylint: disable=W0703
except Exception:
return jsonify(
status=128,
# use better message
message="Metadata file non-existent!")
def compute_gwa_with_loco_covar(k_filename, maf, token):
"""Compute GWA values. No Covariates provided. Only loco and maf vals given.
"""
working_dir = os.path.join(current_app.config.get("TMPDIR"), token)
_dict = jsonfile_to_dict(os.path.join(working_dir, "metadata.json"))
try:
phenofile, snpsfile, covarfile = [
os.path.join(working_dir, _dict.get(x))
for x in ["pheno", "snps", "covar"]
]
genofile = cache_ipfs_file(
ipfs_file=_dict.get("geno"),
cache_dir=current_app.config.get('CACHEDIR'))
if not do_paths_exist([genofile, phenofile, snpsfile, covarfile]):
raise FileNotFoundError
gemma_kwargs = {
"g": genofile,
"p": phenofile,
"a": snpsfile,
"c": covarfile,
"lmm": _dict.get("lmm", 9),
"maf": float(maf)
}
results = generate_gemma_cmd(
gemma_cmd=current_app.config.get("GEMMA_"
"WRAPPER_CMD"),
output_dir=current_app.config.get('TMPDIR'),
token=token,
gemma_kwargs=gemma_kwargs,
gemma_wrapper_kwargs={
"loco": f"--input {os.path.join(working_dir, k_filename)}"
})
return jsonify(unique_id=queue_cmd(
conn=redis.Redis(),
email=(request.get_json() or {}).get('email'),
job_queue=current_app.config.get("REDIS_JOB_QUEUE"),
cmd=results.get("gemma_cmd")),
status="queued",
output_file=results.get("output_file"))
# pylint: disable=W0703
except Exception:
return jsonify(
status=128,
# use better message
message="Metadata file non-existent!")
def test_cache_ipfs_file_cache_hit(self):
"""Test that the correct file location is returned if there's a cache hit"""
# Create empty file
test_dir = "/tmp/QmQPeNsJPyVWPFDVHb77w8G42Fvo15z4bG2X8D2GhfbSXc-test"
if not os.path.exists(test_dir):
os.mkdir(test_dir)
open(f"{test_dir}/genotype.txt", "a").close()
file_loc = cache_ipfs_file(
ipfs_file=("/ipfs/"
"QmQPeNsJPyVWPFDVHb"
"77w8G42Fvo15z4bG2X8D2GhfbSXc-test/"
"genotype.txt"),
cache_dir="/tmp")
# Clean up
os.remove(f"{test_dir}/genotype.txt")
os.rmdir(test_dir)
self.assertEqual(file_loc, f"{test_dir}/genotype.txt")
def compute_gwa(k_filename, token):
"""Compute GWA values. No loco no covariates provided.
"""
working_dir = os.path.join(current_app.config.get("TMPDIR"), token)
_dict = jsonfile_to_dict(os.path.join(working_dir, "metadata.json"))
try:
phenofile, snpsfile = [
os.path.join(working_dir, _dict.get(x)) for x in ["pheno", "snps"]
]
genofile = cache_ipfs_file(
ipfs_file=_dict.get("geno"),
cache_dir=current_app.config.get('CACHEDIR'))
gemma_kwargs = {
"g": genofile,
"p": phenofile,
"a": snpsfile,
"lmm": _dict.get("lmm", 9)
}
results = generate_gemma_cmd(
gemma_cmd=current_app.config.get("GEMMA_"
"WRAPPER_CMD"),
output_dir=current_app.config.get('TMPDIR'),
token=token,
gemma_kwargs=gemma_kwargs,
gemma_wrapper_kwargs={
"input": os.path.join(working_dir, k_filename)
})
return jsonify(unique_id=queue_cmd(
conn=redis.Redis(),
email=(request.get_json() or {}).get('email'),
job_queue=current_app.config.get("REDIS_JOB_QUEUE"),
cmd=results.get("gemma_cmd")),
status="queued",
output_file=results.get("output_file"))
# pylint: disable=W0703
except Exception:
return jsonify(
status=128,
# use better message
message="Metadata file non-existent!")
def test_cache_ipfs_file_cache_miss(self,
mock_ipfs):
"""Test that a file is cached if there's a cache miss"""
mock_ipfs_client = mock.MagicMock()
mock_ipfs.connect.return_value = mock_ipfs_client
test_dir = "/tmp/QmQPeNsJPyVWPFDVHb77w8G42Fvo15z4bG2X8D2GhfbSXc-test"
self.assertEqual(cache_ipfs_file(
ipfs_file=("/ipfs/"
"QmQPeNsJPyVWPFDVHb"
"77w8G42Fvo15z4bG2X8D2GhfbSXc-test/"
"genotype.txt"),
cache_dir="/tmp"
), f"{test_dir}/genotype.txt")
mock_ipfs_client.get.assert_called_once_with(
("/ipfs/"
"QmQPeNsJPyVWPFDVHb"
"77w8G42Fvo15z4bG2X8D2GhfbSXc-test/"
"genotype.txt"),
target=f"{test_dir}"
)
def compute_k_loco(chromosomes, token):
"""Similar to 'compute_k' with the extra option of using loco given chromosome
values.
"""
working_dir = os.path.join(current_app.config.get("TMPDIR"), token)
_dict = jsonfile_to_dict(os.path.join(working_dir, "metadata.json"))
try:
phenofile, snpsfile = [
os.path.join(working_dir, _dict.get(x)) for x in ["pheno", "snps"]
]
genofile = cache_ipfs_file(
ipfs_file=_dict.get("geno"),
cache_dir=current_app.config.get('CACHEDIR'))
if not do_paths_exist([genofile, phenofile, snpsfile]):
raise FileNotFoundError
gemma_kwargs = {"g": genofile, "p": phenofile, "a": snpsfile}
results = generate_gemma_cmd(
gemma_cmd=current_app.config.get("GEMMA_"
"WRAPPER_CMD"),
output_dir=current_app.config.get('TMPDIR'),
token=token,
gemma_kwargs=gemma_kwargs,
chromosomes=chromosomes)
return jsonify(unique_id=queue_cmd(
conn=redis.Redis(),
email=(request.get_json() or {}).get('email'),
job_queue=current_app.config.get("REDIS_JOB_QUEUE"),
cmd=results.get("gemma_cmd")),
status="queued",
output_file=results.get("output_file"))
# pylint: disable=W0703
except Exception:
return jsonify(
status=128,
# use better message
message="Metadata file non-existent!")