Pipeline failure in OPERA-MS v0.9.0 (gap filling + polishing)

[CaroleBelliardo]

Summary

Execution of OPERA-MS (v0.9.0) fails systematically in a Singularity-based HPC environment when processing hybrid PacBio HiFi + Illumina datasets.

Two blocking issues were encountered:

1. ❌ Failure during gap filling step (Step 8)
2. ❌ Failure during polishing step (Pilon) when gap filling is disabled

As a result, the pipeline cannot be completed successfully.

---

🧪 Environment

• Tool version: OPERA-MS v0.9.0
• Container: Singularity (built from Docker)
• HPC system: SLURM
• Resources:
  • 120 CPUs
  • 800 GB RAM

---

📂 Input data

• PacBio HiFi long reads (FASTA)
• Illumina paired-end reads (FASTQ.gz)
• Short-read assembly (FASTA)

---

🚨 Issue 1 — Gap filling failure

⚙️ Command

perl OPERA-MS.pl \
  --num-processors 120 \
  --no-ref-clustering \
  --contig-file "$assembly" \
  --short-read1 "$R1" \
  --short-read2 "$R2" \
  --long-read "$LR" \
  --out-dir "$OUT"

❌ Error

Error in during gap_filling.
Please see gap_filling.err for details.

---

🚨 Issue 2 — Pilon failure (with --no-gap-filling)

⚙️ Command option

--no-gap-filling

❌ Error

Error in during pilon.
Please see pilon.err for details.

---

⚠️

• Pipeline cannot complete under default settings
• Pipeline cannot complete with a workaround
• No final assembly produced

---

❓ Questions

1. Is gap filling known to fail with HiFi reads or high threads?
2. Recommended parameters for gapfilling.pl?
3. Known issues with Singularity vs Docker?

[jsgounot]

Could you share the content of the error files?