---

Whole Genome Sequencing Analysis Pipelines (Nextflow)

Modular short-read and long-read genome assembly, annotation, and taxonomic profiling

---

Project Overview

This repository contains a collection of reproducible Whole Genome Sequencing (WGS) workflows implemented using Nextflow DSL2. The pipelines support:

• Illumina short-read data (single-end and paired-end)
• Long-read data from Oxford Nanopore and PacBio
• Taxonomic profiling (Kraken2), assembly, annotation, and QC

The workflows were developed through hands-on bioinformatics research and pipeline development work at Brigham and Women’s Hospital (BWH), with a focus on real-world usability, modular design, and biological correctness rather than single-dataset optimization.

---

Why This Project Exists

WGS projects often require:

• Different handling for short-read vs long-read technologies
• Reproducible execution across local machines, Conda, and Docker
• Clear separation between workflow logic and compute configuration
• Transparency about biological and data-driven limitations

This project addresses those needs using:

• Nextflow DSL2 for workflow orchestration
• Technology-specific pipelines instead of one “one-size-fits-all” workflow
• Selective containerization (e.g., Docker for Prokka only)
• Explicit documentation of coverage-dependent behavior

---

What Makes This Project Different

• Technology-aware design Separate workflows are provided for Illumina, Nanopore, and PacBio data, reflecting the biological and computational differences between sequencing platforms.

• Modular configuration strategy Compute resources and execution logic are separated into:

  • base.config — shared defaults
  • resources.config — process-level resource tuning
  • nextflow.config — execution profiles (local / conda / docker)

• Selective container usage Only tools that benefit from containerization (e.g., Prokka) run in Docker; all other steps can run via Conda.

• Real-world constraints acknowledged Long-read assemblies are explicitly coverage-dependent. Low-coverage public datasets may fail to assemble complete genomes, and this behavior is documented rather than hidden.

• Grounded in research experience The pipeline design reflects hands-on experience with microbial genome assembly, annotation, and comparative genomics, including a published WGS study.

---

Repository Structure

Whole_Genome_Sequencing_Nextflow/
├── environment_setup.md
├── README.md
├── conf/
│   ├── base.config                  # Shared defaults (CPUs, memory, time, outdir)
│   └── resources.config             # Process-specific resource tuning
├── data/                            # Example input structure (no real data)
│   ├── illumina_PE/
│   │   ├── sample_1.fastq.gz
│   │   └── sample_2.fastq.gz
│   ├── illumina_SE/
│   │   └── sample.fastq.gz
│   ├── nanopore/
│   │   └── sample.fastq.gz
│   ├── pacbio/
│   │   └── sample.fastq.gz
│   └── kraken_library_fasta/
│       └── example_genome.fna
├── nextflow.config                  # Root config with execution profiles
└── workflows/
    ├── Kraken2.nf                   # Taxonomic profiling
    ├── WGS_illumina_PE.nf            # Illumina paired-end pipeline
    ├── WGS_illumina_SE.nf            # Illumina single-end pipeline
    ├── WGS_nanopore.nf               # Oxford Nanopore pipeline
    └── WGS_pacBio.nf                 # PacBio pipeline

---

Workflow Architecture

Short-read (Illumina) workflows

1. Read QC (FastQC)
2. Read trimming (Trimmomatic)
3. Genome assembly (SPAdes)
4. Assembly QC (QUAST)
5. Genome annotation (Prokka)
6. Optional taxonomic classification (Kraken2)

Long-read (Nanopore / PacBio) workflows

1. Taxonomic profiling (Kraken2)
2. User-defined genome size
3. Long-read assembly
4. Genome annotation (Prokka via Docker)
5. Assembly QC (QUAST)

Each workflow can be run independently, depending on sequencing technology and experimental design.

---

How to Run

Note: nextflow.config automatically loads conf/base.config and conf/resources.config. You only need to specify inputs and high-level options at runtime.

Environment Setup

Detailed installation and dependency setup steps for macOS (Apple Silicon) are provided in environment_setup.md.

Illumina Paired-End

nextflow run workflows/WGS_illumina_PE.nf \
  --reads "data/illumina_PE_data/*_{1,2}.fastq.gz" \
  --kraken_library_fasta "kraken_library_fasta" \
  --buildKrakenDB true \
  --kingdom "Bacteria" \
  --genus "Escherichia" \
  --outdir "Results" \
  -profile docker

Illumina Single-End

nextflow run workflows/WGS_illumina_SE.nf \
  --reads "data/illumina_SE_data/*.fastq.gz" \
  --kraken_library_fasta "kraken_library_fasta" \
  --buildKrakenDB true \
  --kingdom "Bacteria" \
  --genus "Escherichia" \
  --outdir "Results" \
  -profile docker

Kraken2 (Quick Species Identification Only)

nextflow run workflows/Kraken2.nf \
  --reads "data/*.fastq.gz" \
  --kraken_library_fasta "kraken_library_fasta" \
  --buildKrakenDB true
  --outdir "Results"

PacBio Long-Read Assembly

nextflow run workflows/WGS_pacBio.nf \
  --reads "data/pacbio_data/*.fastq.gz" \
  --genome_size 4.6m \
  --kingdom "Bacteria" \
  --genus "Escherichia" \
  --species "coli" \
  --outdir "Results" \
  -profile docker

Oxford Nanopore Assembly

nextflow run workflows/WGS_nanopore.nf \
  --reads "data/nanopore_data/*.fastq.gz" \
  --genome_size 4.6m \
  --kingdom "Bacteria" \
  --genus "Escherichia" \
  --species "coli" \
  --outdir "Results" \
  -profile docker

Validation and Known Limitations

The workflows in this repository were validated by running them end-to-end on representative test datasets. The following behaviors were observed during validation and are documented to ensure transparent interpretation of results.

Kraken2 Database Handling

• The pipeline executes successfully when --buildKrakenDB true is used.
  • On the first run, the Kraken2 database is built as expected.
  • On subsequent runs, Nextflow correctly detects the existing database and skips rebuilding, proceeding directly to classification.
• When running with --buildKrakenDB false, Kraken2 classification fails with: despite the required .k2d files being present on disk.
• This behavior was consistently observed during validation for both short-read and long-read workflows.
• Current validated usage: run the workflows with --buildKrakenDB true.

Assembly Quality and QUAST Behavior

• Assembly outcomes depend on sequencing depth and input data quality.
• For some test datasets, assemblies produced only short contigs.
• In these cases, QUAST exited with status 4, consistent with its default minimum contig length requirement (≥500 bp).
• This reflects the biological characteristics of the input data rather than a pipeline execution failure.

Long-read Test Datasets

• Long-read workflows were validated using publicly available test datasets.
• For datasets with insufficient coverage, assemblies did not produce usable contigs, which is expected behavior for long-read assemblers under low-coverage conditions.

Summary

• ✅ Workflows execute end-to-end under validated configurations
• ⚠️ Kraken2 database reuse with --buildKrakenDB false requires refinement
• ⚠️ Assembly and QC outcomes are dependent on sequencing depth and data quality

These observations reflect real-world data constraints and are documented to guide correct usage and future improvements.

---

Relation to Prior Work

This project is informed by:

• Bioinformatics research experience at Brigham and Women’s Hospital

• A published WGS study:

  • Assembly, annotation, and comparative whole genome sequence of Fusarium verticillioides

• Practical experience with Illumina, Nanopore, and PacBio microbial datasets

---

Author

Varun Gowda Bioinformatics | Genomics | Workflow Development <<<<<<< HEAD

---

=======

09d86e2 (Finalize WGS Nextflow pipelines and repository structure)