🧬 OmicsFlow

A modular, containerized NGS pipeline for RNA-seq, long-read, and metagenomic analysis

📋 Overview

OmicsFlow is a production-ready bioinformatics pipeline built with Nextflow and Docker, designed for reproducible multi-omics data analysis. It supports three major sequencing technologies and workflows:

Workflow	Technology	Key tools	Status
rnaseq.nf	Illumina short reads	FastQC · STAR · Salmon · DESeq2	✅ Stable
longread.nf	Oxford Nanopore (ONT)	NanoStat · Minimap2 · Samtools	🚧 In development
metagenomics.nf	Illumina / ONT	Kraken2 · Bracken	🚧 In development

All workflows are fully containerized via Docker and can run locally, on HPC clusters (SLURM/PBS), or in the cloud (AWS Batch).

---

📸 Pipeline Results

Pipeline Architecture

Differential Expression — Volcano Plot

Top 50 DE Genes — Heatmap

Sample Clustering — PCA

---

📊 Validation Metrics

Benchmarked on nf-core test dataset (S. cerevisiae, GSE110004, 4 samples × 50,000 reads):

Metric	Value
Input reads per sample	50,000
Reads passing QC	99.5%
Adapter contamination (auto-detected & removed)	40.3%
Uniquely mapped reads (STAR)	81.8% – 84.6%
Properly paired reads	100%
Mismatch rate	0.9%
Pipeline execution time (4 samples, 4 CPUs)	~8 min
Docker image size	4.63 GB

---

📦 What you need before starting

OmicsFlow is flexible — you can use the full pipeline or individual tools depending on your needs.

The only real requirement: your data

Use case	What you need
Quality control only	FASTQ files
Trimming only	FASTQ files
Alignment (STAR)	FASTQ files + reference genome + GTF + STAR index
Quantification (Salmon)	FASTQ files + Salmon index
Statistics (Samtools)	An existing BAM file
Differential expression	Salmon counts + sample metadata
Python / R analysis	Your own data + scripts

You do not need to prepare everything upfront. Start with what you have and add steps as needed.

---

Reference genome & annotation (only if using STAR alignment)

If you plan to use STAR for alignment, you need a reference genome and its annotation.

If you already have a STAR index on your server or HPC — just point to it with --genomeDir. No need to rebuild it. Any STAR-compatible index works, regardless of how it was generated.

If you need to build one (one-time operation, ~45 min for full human genome):

# Download reference genome (human GRCh38)
wget https://ftp.ensembl.org/pub/release-109/fasta/homo_sapiens/dna/Homo_sapiens.GRCh38.dna.primary_assembly.fa.gz

# Download gene annotation
wget https://ftp.ensembl.org/pub/release-109/gtf/homo_sapiens/Homo_sapiens.GRCh38.109.gtf.gz
gunzip Homo_sapiens.GRCh38.109.gtf.gz

# Build STAR index using OmicsFlow Docker
docker run --rm -v $(pwd):/data smill/omicsflow:1.0.0 \
  bash -c "mkdir -p /data/star_index && STAR --runMode genomeGenerate \
  --genomeDir /data/star_index \
  --genomeFastaFiles /data/genome/GRCh38.fa \
  --sjdbGTFfile /data/genome/GRCh38.gtf \
  --runThreadN 8"

⚠️ Build the index once, store it, reuse it forever for all your experiments.

---

What you do NOT need to install

Everything is already inside the Docker image:

Tool	Without OmicsFlow	With OmicsFlow
FastQC	Manual install	✅ Included
Trim Galore	Manual install	✅ Included
STAR	Compile from source	✅ Included
Salmon	Manual install	✅ Included
Samtools	Compile from source	✅ Included
DESeq2	R + Bioconductor setup	✅ Included
MultiQC	pip install	✅ Included
BioPython	pip install	✅ Included
numpy / pandas / matplotlib	pip install	✅ Included
NanoStat / NanoPlot	pip install	✅ Included
Kraken2	Manual install	✅ Included
Minimap2	Manual install	✅ Included

---

🐳 Run with Docker only (no Nextflow required)

The easiest way to use OmicsFlow — just Docker, no installation needed.

# Pull the image
docker pull smill/omicsflow:1.0.0

# Step 1 — Quality control (FastQC)
docker run --rm -v $(pwd)/data:/data smill/omicsflow:1.0.0 \
  bash -c "fastqc /data/sample_R1.fastq.gz /data/sample_R2.fastq.gz --outdir /data/qc"

# Step 2 — Adapter trimming (Trim Galore)
docker run --rm -v $(pwd)/data:/data smill/omicsflow:1.0.0 \
  bash -c "trim_galore --paired --cores 4 \
  /data/sample_R1.fastq.gz /data/sample_R2.fastq.gz \
  -o /data/trimmed"

# Step 3 — Alignment (STAR)
docker run --rm -v $(pwd)/data:/data smill/omicsflow:1.0.0 \
  bash -c "STAR --runMode alignReads \
  --genomeDir /data/star_index \
  --readFilesIn /data/trimmed/sample_R1_val_1.fq.gz /data/trimmed/sample_R2_val_2.fq.gz \
  --readFilesCommand zcat \
  --outSAMtype BAM SortedByCoordinate \
  --outFileNamePrefix /data/aligned/sample. \
  --runThreadN 4"

# Step 4 — Quantification (Salmon)
docker run --rm -v $(pwd)/data:/data smill/omicsflow:1.0.0 \
  bash -c "salmon quant \
  --index /data/salmon_index \
  --libType A \
  -1 /data/trimmed/sample_R1_val_1.fq.gz \
  -2 /data/trimmed/sample_R2_val_2.fq.gz \
  --output /data/counts/sample \
  --threads 4 \
  --validateMappings"

# Step 5 — BAM statistics (Samtools)
docker run --rm -v $(pwd)/data:/data smill/omicsflow:1.0.0 \
  bash -c "samtools flagstat /data/aligned/sample.Aligned.sortedByCoord.out.bam"

# Step 6 — Aggregated QC report (MultiQC)
docker run --rm -v $(pwd)/data:/data smill/omicsflow:1.0.0 \
  bash -c "multiqc /data --outdir /data/multiqc"

# Interactive R session (DESeq2, ggplot2...)
docker run --rm -it -v $(pwd)/data:/data smill/omicsflow:1.0.0 R

# Interactive Python session (biopython, pandas, matplotlib...)
docker run --rm -it -v $(pwd)/data:/data smill/omicsflow:1.0.0 python3

Windows users: replace $(pwd) with %cd% in CMD, or use the full path.

---

🚀 Run with Nextflow (recommended for production)

Prerequisites

• Nextflow ≥ 22.10
• Docker or Singularity
• Java 17+

Run in one command

# Clone the repository
git clone https://github.com/Millimono/OmicsFlow.git
cd OmicsFlow

# Run RNA-seq pipeline with test data
nextflow run workflows/rnaseq.nf \
  --input data/test/samplesheet.csv \
  --genome GRCh38 \
  --outdir results/ \
  -profile docker

Input samplesheet format (CSV)

sample,fastq_1,fastq_2,strandedness
ctrl_rep1,/path/to/ctrl_rep1_R1.fastq.gz,/path/to/ctrl_rep1_R2.fastq.gz,reverse
ctrl_rep2,/path/to/ctrl_rep2_R1.fastq.gz,/path/to/ctrl_rep2_R2.fastq.gz,reverse
treat_rep1,/path/to/treat_rep1_R1.fastq.gz,/path/to/treat_rep1_R2.fastq.gz,reverse
treat_rep2,/path/to/treat_rep2_R1.fastq.gz,/path/to/treat_rep2_R2.fastq.gz,reverse

Strandedness: use reverse for most Illumina TruSeq kits, forward for some stranded protocols, unstranded if unsure.

---

🗂️ Project Structure

OmicsFlow/
├── workflows/
│   ├── rnaseq.nf           # ✅ RNA-seq Illumina pipeline (stable)
│   ├── longread.nf         # 🚧 Nanopore long-read pipeline (in development)
│   └── metagenomics.nf     # 🚧 Metagenomic pipeline (in development)
│
├── modules/
│   ├── qc/                 # FastQC, MultiQC, NanoStat
│   ├── alignment/          # STAR, Minimap2, Samtools
│   └── quantification/     # Salmon, DESeq2
│
├── analysis/
│   ├── deseq2.R            # Differential expression (DESeq2 / edgeR)
│   ├── plots.py            # Heatmaps, volcano plots, PCA
│   └── report.Rmd          # Automated HTML report template
│
├── containers/
│   └── Dockerfile          # All tools in one reproducible image
│
├── data/
│   └── test/               # Public mini-datasets for testing
│       ├── samplesheet.csv
│       └── reads/          # nf-core GSE110004 subset (Illumina)
│
├── docs/                   # Documentation (GitHub Pages)
├── .github/
│   └── workflows/
│       └── ci.yml          # GitHub Actions CI/CD
└── nextflow.config         # Profiles: local, cluster, cloud

---

📊 Workflows in Detail

1. RNA-seq Pipeline (rnaseq.nf) ✅ Stable

Designed for bulk RNA-seq analysis from raw FASTQ to differential expression.

Input FASTQ
    │
    ▼
[FastQC] ──────────────────────────> QC report
    │
    ▼
[Trim Galore] ──> Trimmed reads
    │
    ▼
[STAR] ──> Aligned BAM + splice junctions
    │
    ▼
[Salmon] ──> Gene/transcript counts
    │
    ▼
[DESeq2 / edgeR] ──> Differential expression
    │
    ▼
[MultiQC] ──> Aggregated QC report (HTML)

Output files:

• results/qc/ — FastQC + MultiQC reports
• results/aligned/ — BAM files + index
• results/counts/ — Salmon quantification
• results/deseq2/ — DE results, volcano plots, heatmaps
• results/report.html — Full automated HTML report

---

2. Long-read Pipeline (longread.nf) 🚧 In development

For Nanopore sequencing data. Coming soon — tools already available in the Docker image.

Input FASTQ (ONT)
    │
    ▼
[NanoStat / NanoPlot] ──> Read quality stats
    │
    ▼
[Minimap2] ──> Aligned BAM
    │
    ▼
[Samtools] ──> Sorted + indexed BAM
    │
    ▼
[MultiQC] ──> Aggregated report

In the meantime, you can use these tools individually via Docker — see the Docker section above.

---

3. Metagenomic Pipeline (metagenomics.nf) 🚧 In development

Taxonomic classification and abundance profiling. Coming soon — Kraken2 already available in the Docker image.

Input FASTQ
    │
    ▼
[FastQC + Trim Galore] ──> Clean reads
    │
    ▼
[Kraken2] ──> Taxonomic classification
    │
    ▼
[Bracken] ──> Abundance re-estimation

In the meantime, you can run Kraken2 directly:

docker run --rm -v $(pwd):/data smill/omicsflow:1.0.0 \
  bash -c "kraken2 --db /data/kraken2_db --paired \
  /data/R1.fastq.gz /data/R2.fastq.gz \
  --output /data/kraken2_output.txt \
  --report /data/kraken2_report.txt"

---

🛠️ Technical Stack

Category	Tools	Versions
Pipeline orchestration	Nextflow DSL2	≥ 22.10
Containerization	Docker · Singularity	28.x
QC	FastQC · MultiQC · NanoStat · NanoPlot	0.12.1 · 1.35 · 1.6.0
Alignment	STAR · Minimap2	2.7.11b · 2.31
Quantification	Salmon	1.12.0
Variant calling	Samtools · BCFtools	1.23.1
Metagenomics	Kraken2	2.1.3
Statistical analysis	DESeq2 · edgeR · R	R 4.5.2
Visualization	ggplot2 · matplotlib · seaborn	—
Languages	Python · R · Bash · C · C++	Python 3.x
CI/CD	GitHub Actions	—
Documentation	GitHub Pages	—

---

🧪 Test Data

Test data used during development (publicly available):

Dataset	Source	Size	Used for
GSE110004 / SRR6357070-71 (4 samples)	nf-core test datasets	~8 MB	RNA-seq validation
S. cerevisiae R64-1-1 genome	nf-core test datasets	~230 KB	Reference genome
S. cerevisiae gene annotation	nf-core test datasets	~200 KB	Gene annotation

---

⚙️ Configuration

OmicsFlow supports multiple execution profiles defined in nextflow.config:

profiles {
    docker {
        docker.enabled   = true
        process.executor = 'local'
    }
    cluster {
        process.executor    = 'slurm'
        singularity.enabled = true
        process.queue       = 'normal'
    }
    cloud {
        process.executor = 'awsbatch'
        aws.region       = 'ca-central-1'
    }
    test {
        params.input  = "${projectDir}/data/test/samplesheet.csv"
        params.outdir = 'results_test'
        docker.enabled = true
    }
}

---

📈 Results & Outputs

Every run generates a timestamped output directory:

results/
├── qc/
│   ├── fastqc/             # Per-sample FastQC reports (HTML)
│   └── multiqc_report.html # Aggregated QC report
├── trimmed/
│   └── logs/               # Trim Galore trimming reports
├── aligned/
│   ├── sample.Aligned.sortedByCoord.out.bam
│   └── sample.Log.final.out  # Mapping statistics
├── counts/
│   └── salmon/             # Transcript-level quantification
│       └── quant.sf
├── deseq2/
│   ├── deseq2_results.csv  # DE genes table
│   ├── volcano_plot.pdf    # Volcano plot
│   ├── heatmap_top50.pdf   # Top 50 DE genes heatmap
│   └── pca_plot.pdf        # PCA plot
└── pipeline_info/
    ├── execution_report.html
    └── execution_timeline.html

---

🔗 Link to Research

This pipeline was developed in conjunction with research in AI-based medical imaging and bioinformatics:

• MalariaScan — AI detection of malaria via blood microscopy. Prix Coup de cœur Jean-Marc Léger, UdeM 2025.
• HAtt-CNN — Adaptive visual attention supervision with heuristic masks for CNN interpretability. (Under review 2026)
• EpitopeNet — Backpropagation-free prototype learning inspired by B-cell dynamics for mammography classification. 76.03% accuracy on MiniDDSM. (Under review 2026)

The analysis/ module is extensible — ML models from the above projects can be integrated as additional pipeline steps.

---

📚 Documentation

Full documentation available at: millimono.github.io/OmicsFlow

---

🤝 Contributing

Contributions welcome! Please open a pull request.

git clone https://github.com/Millimono/OmicsFlow.git
cd OmicsFlow
git checkout -b feature/my-new-module

---

📄 Citation

Millimono, S. (2026). OmicsFlow: A modular containerized NGS pipeline 
for reproducible multi-omics analysis (v1.0.1). Zenodo. 
https://doi.org/10.5281/zenodo.20677900

---

👤 Author

Sory Millimono PhD Candidate in AI · Bioinformatician Université de Montréal & Mohammed V University – ENSIAS

• 📧 millimono64.sm@gmail.com
• 🔗 LinkedIn
• 🎓 Google Scholar — h-index 1 · 24 citations
• 🔬 ORCID: 0009-0005-1960-9136

---

📜 License

MIT License — see LICENSE for details.