Fix stage_data.nf: use synthea:cudf container, Python-based downloads with sshpass/expect fallback
This commit is contained in:
139
stage_data.nf
139
stage_data.nf
@@ -1,83 +1,104 @@
|
||||
nextflow.enable.dsl=2
|
||||
|
||||
process STAGE_DATA {
|
||||
container 'ubuntu:22.04'
|
||||
container 'harbor.cluster.omic.ai/omic/digital-patients/synthea:cudf'
|
||||
memory '4 GB'
|
||||
cpus 2
|
||||
|
||||
script:
|
||||
"""
|
||||
#!/opt/conda/envs/synthea/bin/python3
|
||||
import subprocess, os, sys
|
||||
|
||||
BASE = "/omic/eureka/digital-patients"
|
||||
|
||||
echo "=== Installing dependencies ==="
|
||||
apt-get update -qq && apt-get install -y -qq openssh-client sshpass curl > /dev/null 2>&1
|
||||
def run(cmd):
|
||||
print(f"RUN: {cmd}", flush=True)
|
||||
r = subprocess.run(cmd, shell=True, capture_output=True, text=True)
|
||||
if r.stdout: print(r.stdout, flush=True)
|
||||
if r.stderr: print(r.stderr, flush=True)
|
||||
return r.returncode
|
||||
|
||||
echo "=== Cleaning up old _parts directories ==="
|
||||
find \$BASE -name "*_parts" -type d -exec rm -rf {} + 2>/dev/null || true
|
||||
echo "Cleanup done"
|
||||
print("=== Cleaning up old _parts directories ===", flush=True)
|
||||
run(f"find {BASE} -name '*_parts' -type d -exec rm -rf {{}} + 2>/dev/null || true")
|
||||
|
||||
echo "=== Creating directory structure ==="
|
||||
mkdir -p \$BASE/imputed \$BASE/healthy \$BASE/supporting-data/vcf \$BASE/supporting-data/ucsc-liftover \$BASE/supporting-data/genome \$BASE/output
|
||||
print("=== Creating directory structure ===", flush=True)
|
||||
for d in ["imputed", "healthy", "supporting-data/vcf", "supporting-data/ucsc-liftover", "supporting-data/genome", "output"]:
|
||||
os.makedirs(f"{BASE}/{d}", exist_ok=True)
|
||||
|
||||
echo "=== Current state of PVC ==="
|
||||
ls -la \$BASE/ 2>/dev/null
|
||||
print("=== Current PVC state ===", flush=True)
|
||||
run(f"ls -la {BASE}/")
|
||||
|
||||
download() {
|
||||
local remote="\$1"
|
||||
local local_path="\$2"
|
||||
if [ -f "\$local_path" ]; then
|
||||
echo "SKIP (exists): \$local_path"
|
||||
ls -lh "\$local_path"
|
||||
return
|
||||
fi
|
||||
echo "Downloading: \$remote -> \$local_path"
|
||||
sshpass -p 'bl3rg3r5' scp -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null -P 9100 "omic@nucleus.omic.ai:\$remote" "\$local_path"
|
||||
ls -lh "\$local_path"
|
||||
}
|
||||
print("=== Installing sshpass ===", flush=True)
|
||||
# Try conda first, then apt
|
||||
rc = run("conda install -y -c conda-forge sshpass 2>/dev/null || apt-get update -qq && apt-get install -y -qq sshpass 2>/dev/null || pip install sshpass 2>/dev/null")
|
||||
# Check if sshpass is available
|
||||
rc = run("which sshpass || echo 'sshpass not found, trying expect'")
|
||||
|
||||
echo "=== Downloading imputed files ==="
|
||||
download "/mnt/Avatar/imputed/ukbb/imputed/F5_SCHIZO.gwas.imputed_v3.both_sexes.tsv.bgz" "\$BASE/imputed/F5_SCHIZO.gwas.imputed_v3.both_sexes.tsv.bgz"
|
||||
download "/mnt/Avatar/imputed/ukbb/imputed/F5_SCHIZO.gwas.imputed_v3.female.tsv.bgz" "\$BASE/imputed/F5_SCHIZO.gwas.imputed_v3.female.tsv.bgz"
|
||||
download "/mnt/Avatar/imputed/ukbb/imputed/F5_SCHIZO.gwas.imputed_v3.male.tsv.bgz" "\$BASE/imputed/F5_SCHIZO.gwas.imputed_v3.male.tsv.bgz"
|
||||
# If sshpass not available, use expect or python paramiko
|
||||
has_sshpass = subprocess.run("which sshpass", shell=True, capture_output=True).returncode == 0
|
||||
|
||||
echo "=== Downloading gnomad files ==="
|
||||
download "/mnt/Avatar/digital_patient/gnomad.genomes.v4.1.sites.female.txt" "\$BASE/healthy/gnomad.genomes.v4.1.sites.female.txt"
|
||||
download "/mnt/Avatar/digital_patient/gnomad.genomes.v4.1.sites.male.txt" "\$BASE/healthy/gnomad.genomes.v4.1.sites.male.txt"
|
||||
def download(remote, local_path):
|
||||
if os.path.isfile(local_path):
|
||||
size = os.path.getsize(local_path)
|
||||
print(f"SKIP (exists, {size} bytes): {local_path}", flush=True)
|
||||
return True
|
||||
print(f"Downloading: {remote} -> {local_path}", flush=True)
|
||||
if has_sshpass:
|
||||
cmd = f"sshpass -p 'bl3rg3r5' scp -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null -P 9100 'omic@nucleus.omic.ai:{remote}' '{local_path}'"
|
||||
else:
|
||||
# Fallback: use expect
|
||||
cmd = f'''expect -c "
|
||||
set timeout 3600
|
||||
spawn scp -o StrictHostKeyChecking=no -P 9100 omic@nucleus.omic.ai:{remote} {local_path}
|
||||
expect \\"password:\\"
|
||||
send \\"bl3rg3r5\\r\\"
|
||||
expect eof
|
||||
"'''
|
||||
rc = run(cmd)
|
||||
if rc == 0 and os.path.isfile(local_path):
|
||||
size = os.path.getsize(local_path)
|
||||
print(f" OK: {size} bytes", flush=True)
|
||||
return True
|
||||
else:
|
||||
print(f" FAILED (rc={rc})", flush=True)
|
||||
return False
|
||||
|
||||
echo "=== Downloading supporting-data files ==="
|
||||
download "/mnt/Avatar/dd/synthea/supporting-data/genome/hg38.fa" "\$BASE/supporting-data/genome/hg38.fa"
|
||||
download "/mnt/Avatar/dd/synthea/supporting-data/genome/hg38.dict" "\$BASE/supporting-data/genome/hg38.dict"
|
||||
download "/mnt/Avatar/dd/synthea/supporting-data/ucsc-liftover/hg19ToHg38.over.chain.gz" "\$BASE/supporting-data/ucsc-liftover/hg19ToHg38.over.chain.gz"
|
||||
download "/mnt/Avatar/dd/synthea/supporting-data/vcf/vcf_template.vcf" "\$BASE/supporting-data/vcf/vcf_template.vcf"
|
||||
files = [
|
||||
("/mnt/Avatar/imputed/ukbb/imputed/F5_SCHIZO.gwas.imputed_v3.both_sexes.tsv.bgz", f"{BASE}/imputed/F5_SCHIZO.gwas.imputed_v3.both_sexes.tsv.bgz"),
|
||||
("/mnt/Avatar/imputed/ukbb/imputed/F5_SCHIZO.gwas.imputed_v3.female.tsv.bgz", f"{BASE}/imputed/F5_SCHIZO.gwas.imputed_v3.female.tsv.bgz"),
|
||||
("/mnt/Avatar/imputed/ukbb/imputed/F5_SCHIZO.gwas.imputed_v3.male.tsv.bgz", f"{BASE}/imputed/F5_SCHIZO.gwas.imputed_v3.male.tsv.bgz"),
|
||||
("/mnt/Avatar/digital_patient/gnomad.genomes.v4.1.sites.female.txt", f"{BASE}/healthy/gnomad.genomes.v4.1.sites.female.txt"),
|
||||
("/mnt/Avatar/digital_patient/gnomad.genomes.v4.1.sites.male.txt", f"{BASE}/healthy/gnomad.genomes.v4.1.sites.male.txt"),
|
||||
("/mnt/Avatar/dd/synthea/supporting-data/genome/hg38.fa", f"{BASE}/supporting-data/genome/hg38.fa"),
|
||||
("/mnt/Avatar/dd/synthea/supporting-data/genome/hg38.dict", f"{BASE}/supporting-data/genome/hg38.dict"),
|
||||
("/mnt/Avatar/dd/synthea/supporting-data/ucsc-liftover/hg19ToHg38.over.chain.gz", f"{BASE}/supporting-data/ucsc-liftover/hg19ToHg38.over.chain.gz"),
|
||||
("/mnt/Avatar/dd/synthea/supporting-data/vcf/vcf_template.vcf", f"{BASE}/supporting-data/vcf/vcf_template.vcf"),
|
||||
("/mnt/Avatar/imputed/ukbb/metadata/ukbb_phenotypes.csv", f"{BASE}/ukbb_phenotypes_filtered.csv"),
|
||||
]
|
||||
|
||||
echo "=== Downloading phenotype file ==="
|
||||
download "/mnt/Avatar/imputed/ukbb/metadata/ukbb_phenotypes.csv" "\$BASE/ukbb_phenotypes_filtered.csv"
|
||||
print("=== Downloading files ===", flush=True)
|
||||
ok = 0
|
||||
fail = 0
|
||||
for remote, local in files:
|
||||
if download(remote, local):
|
||||
ok += 1
|
||||
else:
|
||||
fail += 1
|
||||
|
||||
echo "=== Copying small files from DRS (already on PVC via DRS upload) ==="
|
||||
# These were uploaded via DRS and should already be on PVC
|
||||
# If not, they're in the git repo that WES cloned
|
||||
for f in MANE.GRCh38.v1.3.update.tsv regulon.rda LM22_sourceGEP_ensg.txt; do
|
||||
if [ ! -f "\$BASE/\$f" ]; then
|
||||
echo "Small file missing on PVC, checking if available from WES workdir..."
|
||||
# WES clones the repo, so the file might be in the current workdir's repo
|
||||
else
|
||||
echo "SKIP (exists): \$BASE/\$f"
|
||||
fi
|
||||
done
|
||||
print(f"\\n=== Download results: {ok} ok, {fail} failed ===", flush=True)
|
||||
|
||||
echo "=== Final verification ==="
|
||||
echo "Imputed:"
|
||||
ls -lh \$BASE/imputed/*.bgz 2>/dev/null || echo " NONE"
|
||||
echo "Healthy:"
|
||||
ls -lh \$BASE/healthy/*.txt 2>/dev/null || echo " NONE"
|
||||
echo "Genome:"
|
||||
ls -lh \$BASE/supporting-data/genome/* 2>/dev/null || echo " NONE"
|
||||
echo "Supporting-data:"
|
||||
ls -lh \$BASE/supporting-data/vcf/* \$BASE/supporting-data/ucsc-liftover/* 2>/dev/null || echo " NONE"
|
||||
echo "Small files:"
|
||||
ls -lh \$BASE/MANE* \$BASE/regulon* \$BASE/LM22* \$BASE/ukbb* 2>/dev/null || echo " NONE"
|
||||
print("\\n=== Final verification ===", flush=True)
|
||||
run(f"echo 'Imputed:' && ls -lh {BASE}/imputed/*.bgz 2>/dev/null || echo ' NONE'")
|
||||
run(f"echo 'Healthy:' && ls -lh {BASE}/healthy/*.txt 2>/dev/null || echo ' NONE'")
|
||||
run(f"echo 'Genome:' && ls -lh {BASE}/supporting-data/genome/* 2>/dev/null || echo ' NONE'")
|
||||
run(f"echo 'Support:' && ls -lh {BASE}/supporting-data/vcf/* {BASE}/supporting-data/ucsc-liftover/* 2>/dev/null || echo ' NONE'")
|
||||
run(f"echo 'Small:' && ls -lh {BASE}/MANE* {BASE}/regulon* {BASE}/LM22* {BASE}/ukbb* 2>/dev/null || echo ' NONE'")
|
||||
|
||||
echo "=== STAGING COMPLETE ==="
|
||||
if fail > 0:
|
||||
print("STAGING FAILED - some files missing", flush=True)
|
||||
sys.exit(1)
|
||||
print("=== STAGING COMPLETE ===", flush=True)
|
||||
"""
|
||||
}
|
||||
|
||||
|
||||
Reference in New Issue
Block a user