omic/digital-patients
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Fix stage_data.nf: pure bash with conda install sshpass, set +e to avoid early exit

Browse Source
This commit is contained in:
Olamide Isreal
2026-04-01 10:27:46 +01:00
parent c3c073e964
commit 8695fa01e9
1 changed files with 56 additions and 81 deletions
Download Patch File Download Diff File Expand all files Collapse all files

137
stage_data.nf
View File

@@ -7,98 +7,73 @@ process STAGE_DATA {
script: script:
""" """
#!/opt/conda/envs/synthea/bin/python3 set +e
import subprocess, os, sys BASE="/omic/eureka/digital-patients"
BASE = "/omic/eureka/digital-patients" echo "=== Cleaning up old _parts directories ==="
find \$BASE -name "*_parts" -type d -exec rm -rf {} + 2>/dev/null
echo "Cleanup done"
def run(cmd): echo "=== Creating directory structure ==="
print(f"RUN: {cmd}", flush=True) mkdir -p \$BASE/imputed \$BASE/healthy \$BASE/supporting-data/vcf \$BASE/supporting-data/ucsc-liftover \$BASE/supporting-data/genome \$BASE/output
r = subprocess.run(cmd, shell=True, capture_output=True, text=True) echo "Dirs created"
if r.stdout: print(r.stdout, flush=True)
if r.stderr: print(r.stderr, flush=True)
return r.returncode
print("=== Cleaning up old _parts directories ===", flush=True) echo "=== Current PVC state ==="
run(f"find {BASE} -name '*_parts' -type d -exec rm -rf {{}} + 2>/dev/null || true") ls -la \$BASE/ 2>/dev/null || echo "Base dir issue"
print("=== Creating directory structure ===", flush=True) echo "=== Installing sshpass ==="
for d in ["imputed", "healthy", "supporting-data/vcf", "supporting-data/ucsc-liftover", "supporting-data/genome", "output"]: conda install -y -c conda-forge sshpass 2>&1 | tail -3
os.makedirs(f"{BASE}/{d}", exist_ok=True) which sshpass && echo "sshpass OK" || echo "sshpass NOT FOUND"
print("=== Current PVC state ===", flush=True) download() {
run(f"ls -la {BASE}/") remote="\$1"
local_path="\$2"
if [ -f "\$local_path" ]; then
echo "SKIP (exists): \$local_path"
return 0
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" 2>&1
if [ -f "\$local_path" ]; then
ls -lh "\$local_path"
return 0
else
echo "FAILED: \$local_path"
return 1
fi
}
print("=== Installing sshpass ===", flush=True) echo "=== Downloading imputed files ==="
# Try conda first, then apt 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"
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") download "/mnt/Avatar/imputed/ukbb/imputed/F5_SCHIZO.gwas.imputed_v3.female.tsv.bgz" "\$BASE/imputed/F5_SCHIZO.gwas.imputed_v3.female.tsv.bgz"
# Check if sshpass is available download "/mnt/Avatar/imputed/ukbb/imputed/F5_SCHIZO.gwas.imputed_v3.male.tsv.bgz" "\$BASE/imputed/F5_SCHIZO.gwas.imputed_v3.male.tsv.bgz"
rc = run("which sshpass || echo 'sshpass not found, trying expect'")
# If sshpass not available, use expect or python paramiko
has_sshpass = subprocess.run("which sshpass", shell=True, capture_output=True).returncode == 0
def download(remote, local_path): echo "=== Downloading gnomad files ==="
if os.path.isfile(local_path): download "/mnt/Avatar/digital_patient/gnomad.genomes.v4.1.sites.female.txt" "\$BASE/healthy/gnomad.genomes.v4.1.sites.female.txt"
size = os.path.getsize(local_path) download "/mnt/Avatar/digital_patient/gnomad.genomes.v4.1.sites.male.txt" "\$BASE/healthy/gnomad.genomes.v4.1.sites.male.txt"
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
files = [ echo "=== Downloading supporting-data 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"), download "/mnt/Avatar/dd/synthea/supporting-data/genome/hg38.fa" "\$BASE/supporting-data/genome/hg38.fa"
("/mnt/Avatar/imputed/ukbb/imputed/F5_SCHIZO.gwas.imputed_v3.female.tsv.bgz", f"{BASE}/imputed/F5_SCHIZO.gwas.imputed_v3.female.tsv.bgz"), download "/mnt/Avatar/dd/synthea/supporting-data/genome/hg38.dict" "\$BASE/supporting-data/genome/hg38.dict"
("/mnt/Avatar/imputed/ukbb/imputed/F5_SCHIZO.gwas.imputed_v3.male.tsv.bgz", f"{BASE}/imputed/F5_SCHIZO.gwas.imputed_v3.male.tsv.bgz"), download "/mnt/Avatar/dd/synthea/supporting-data/ucsc-liftover/hg19ToHg38.over.chain.gz" "\$BASE/supporting-data/ucsc-liftover/hg19ToHg38.over.chain.gz"
("/mnt/Avatar/digital_patient/gnomad.genomes.v4.1.sites.female.txt", f"{BASE}/healthy/gnomad.genomes.v4.1.sites.female.txt"), download "/mnt/Avatar/dd/synthea/supporting-data/vcf/vcf_template.vcf" "\$BASE/supporting-data/vcf/vcf_template.vcf"
("/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"),
]
print("=== Downloading files ===", flush=True) echo "=== Downloading phenotype file ==="
ok = 0 download "/mnt/Avatar/imputed/ukbb/metadata/ukbb_phenotypes.csv" "\$BASE/ukbb_phenotypes_filtered.csv"
fail = 0
for remote, local in files:
if download(remote, local):
ok += 1
else:
fail += 1
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 "Support:"
ls -lh \$BASE/supporting-data/vcf/* \$BASE/supporting-data/ucsc-liftover/* 2>/dev/null || echo " NONE"
echo "Small:"
ls -lh \$BASE/MANE* \$BASE/regulon* \$BASE/LM22* \$BASE/ukbb* 2>/dev/null || echo " NONE"
print("\\n=== Final verification ===", flush=True) echo "=== STAGING COMPLETE ==="
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'")
if fail > 0:
print("STAGING FAILED - some files missing", flush=True)
sys.exit(1)
print("=== STAGING COMPLETE ===", flush=True)
""" """
} }