Viral Metagenomics Pipeline Development from HPC Submission to Monitoring, and Interactive Analysis

Viral Metagenomics Pipeline Development from HPC Submission to Monitoring, and Interactive Analysis

Description

Context

The CIRI is a research unit (INSERM, CNRS, Université Lyon 1, ENS de Lyon) spanning Virology, Bacteriology, and Immunology.

This internship is a collaboration between the Ibiv team (Dr. Olivier Reynard) and the Bioinformatics Hub (Dr. Julien Fouret). The goal is to deploy a reusable web interface enabling biologists to submit HPC workloads on the local PSMN-ENS Lyon infrastructure, centered on a viral metagenomics pipeline and complemented by in-house exploration tools.

The Ibiv team studies emerging zoonotic viruses , notably bat-associated Paramyxoviridae (Henipaviruses). Dr. Reynard recently received ANR funding to investigate Parahenipaviruses identified in French shrews — close relatives of bat viruses. Objectives: characterize genetic diversity, assess prevalence in wild populations, and identify viruses with pathogenic potential, all essential for zoonotic preparedness .

The Bioinformatics Hub’s mission is to democratize bioinformatics across CIRI teams. With limited human resources, it invests in training and interactive tool deployment. This internship delivers a solution tailored to Dr. Reynard’s needs, designed for reuse across teams. The hub has already initiated a Flask web app nf-web and selected and integrated the nf-core/viralmetagenome pipeline.

Co-supervision: Olivier Reynard (product owner) / Julien Fouret (technical lead).

Missions

1. Explore and valorize the nf-core/viralmetagenome outputs

Understand the full analytical output: QC reads, host-depleted reads, de novo and reference-guided viral assemblies , contig taxonomic classification (Kraken2/BLAST), read mapping statistics, consensus sequences.

Build an interactive exploration interface (Shiny or Dash) providing:

• Interactive filtering of BLAST results (reads and contigs)
• Taxonomic composition plots (Krona-like, Sankey)
• Assembly quality metrics (N50, GC content, CheckV completeness)

2. Complementary downstream analyses and extension workflows

Implement a secondary analysis pipeline (Nextflow or post-processing modules):

• Terminal-sequence re-assembly via iterative mapping — refining genome ends by successive read-mapping/extension rounds, critical for paramyxovirus leader/trailer regions missed by standard assemblers
• Targeted reannotation using curated paramyxovirus protein databases (P/V/C, F, HN, N, L) to refine gene boundaries and detect P-gene editing sites, alternative ORFs, and F protein cleavage motifs

3. Extend the web application for HPC submission

The existing Flask app already provides:

• Workflow selection with version pinning
• HPC configuration (partition, CPU/memory/time, Singularity/Apptainer container paths)
• Dynamic parameterization via JSON schema-driven forms (nf-schema): sample sheets, reference genomes, database indices, with data-portal file picker links

You will extend it with:

• Run launching with pre-flight validation (schema compliance, input existence) and run resume support
• Real-time run management : streaming Nextflow logs, per-task resource metrics, failure diagnostics
• Run dashboard : summary statistics and direct links to outputs

Data transfer: integrate a PSMN-hosted solution or a FileZilla-based approach for uploading/downloading sequencing files (fastq/fasta) and metadata tables.

Desired skills & knowledge

All the stack is not required, the intern will be trained on what he is not familiar with.

• Viral genomics: read preprocessing (trimming, QC, host depletion), assembly challenges (repeats, circular genomes, low termini coverage), mapping and consensus (minimap2, Bowtie2, iVar, bcftools)
• HPC & workflows: SLURM, SSH remote work, Nextflow in HPC, Singularity/Apptainer
• DevOps: Git; Ansible and CI/CD exposure appreciated
• Interactive visualization: R Shiny or Python Dash/Streamlit for genomic data
• UI/UX: intuitive interfaces for non-bioinformaticians; form logic, error messaging, data provenance