Nextflow DSL2 · Docker Ready · v1.0.0-beta

Clinical Bacterial WGS
from raw reads to report

A reproducible, containerized Nextflow pipeline for clinical microbiology labs — automating QC, assembly, AMR typing, MLST, and phylogenetic surveillance into a single audit-ready HTML report.

nextflow ≥23.10 DSL2 Docker Singularity MIT License CI: passing

$ nextflow run your-github/BacTrack-NF -profile docker

The Problem

Clinical labs need a pipeline
that actually fits their workflow

Whole-genome sequencing is now standard in clinical microbiology. But the bioinformatics tooling hasn't kept up — labs stitch together ad-hoc scripts that break, can't be audited, and aren't reproducible between runs.

Current Pain Points

Existing tools (Bactopia, Nullarbor) are research-focused, not clinical-grade
No single pipeline covers QC → assembly → AMR → MLST → phylogeny end-to-end
Outputs aren't audit-ready — no structured HTML report for clinical record-keeping
Hard to deploy in air-gapped hospital HPC environments
No built-in outbreak detection or cluster alert thresholds

BacTrack-NF solves this

Full DSL2 Nextflow pipeline with modular, testable processes
Six integrated modules: QC → Assembly → Annotation → AMR → MLST → Phylogeny
Structured HTML + JSON report suitable for clinical audit trails
Singularity support for air-gapped HPC; all containers pre-pulled and versioned
SNP-distance outbreak alerting with configurable cluster thresholds

Architecture

Six modules, one command

Each module is a self-contained Nextflow process with its own Docker container, test data, and documentation. Mix and match with --skip_* flags.

📥

Input QC

FastQC

fastp

🧩

Assembly

Shovill

SPAdes

🔬

Annotation

Prokka

QUAST

💊

AMR Detection

AMRFinder+

RESFinder

🧬

MLST Typing

mlst

Kleborate

🌳

Phylogeny

Snippy

FastTree2

Key Features

Built for clinical reproducibility

📋

Audit-Ready HTML Reports

Every run produces a structured, timestamped HTML + JSON report with QC flags, AMR profiles, and chain-of-custody metadata for clinical records.

report.html · report.json

🚨

Outbreak Cluster Alerting

SNP-distance matrix with configurable threshold alerts. Isolates within N SNPs of a reference are flagged automatically — critical for infection control.

--snp_threshold 10

🐳

Air-Gap & HPC Ready

All containers are versioned and can be pre-pulled. Singularity profile included for hospital HPCs with no internet access. Zero dependency hell.

-profile singularity

🧪

CI-Tested with nf-test

Each module ships with nf-test unit tests run on every commit. Snapshot testing ensures outputs remain consistent across tool version updates.

nf-test · GitHub Actions

⚙️

Flexible Config Profiles

Pre-built profiles for local laptop, Slurm HPC, AWS Batch, and Google Cloud. Switch with a single flag — no config file editing required.

-profile slurm,docker

📊

MultiQC Integration

Aggregate QC across all samples in a run. Instantly flag low-coverage, contaminated, or failed samples before downstream analysis wastes compute time.

multiqc · coverage · N50

Example Output

Mock results from 12 K. pneumoniae isolates

Simulated run on clinical-grade Illumina reads (2×150bp, ~100× coverage). All outputs generated by the pipeline with no manual intervention.

12/12

Samples passed QC

5.4 Mb

Mean assembly size

98.2×

Mean coverage depth

127 kb

Mean N50

Mean Q30 Score

94.1%

Adapter trimmed

8.2%

Mapping rate

99.3%

Contigs < 500 bp

3.8%

Sample ID	Gene	Drug Class	% Identity	Coverage	Phenotype
KP-001	blaCTX-M-15	Beta-lactam (ESBL)	99.8%	100%	R
KP-001	aac(6')-Ib-cr	Aminoglycoside	98.2%	100%	R
KP-003	oqxAB	Fluoroquinolone	97.5%	98%	I
KP-005	blaTEM-1B	Beta-lactam	100%	100%	R
KP-007	mcr-1.1	Colistin	99.1%	99%	R
KP-009	— (no AMR genes)	—	—	—	S
KP-011	blaKPC-2	Carbapenem	100%	100%	R

Klebsiella pneumoniae

ST258

gapA:2infB:1mdh:1 pgi:1phoE:9rpoB:4tonB:12

Klebsiella pneumoniae

ST307

gapA:2infB:1mdh:1 pgi:1phoE:1rpoB:1tonB:70

Klebsiella pneumoniae

ST147

gapA:3infB:3mdh:1 pgi:1phoE:8rpoB:4tonB:4

Klebsiella pneumoniae

ST11

gapA:2infB:2mdh:2 pgi:2phoE:3rpoB:2tonB:7

Getting Started

Up and running in three commands

# 1. Install Nextflow (requires Java 17+)
curl -s https://get.nextflow.io | bash

# 2. Pull and run BacTrack-NF
nextflow run your-github/BacTrack-NF \
  --input samplesheet.csv \
  --outdir ./results \
  --genome GCF_000240185.1 \
  --snp_threshold 10 \
  -profile docker

# 3. Open results/report.html in any browser
open results/report.html
    

// nextflow.config — customise resources per process
params {
  input          = null
  outdir         = "results"
  genome         = null
  snp_threshold  = 20
  min_coverage   = 30
  skip_annotation= false
  skip_phylogeny = false
}

process {
  withName: 'SHOVILL' {
    cpus   = 8
    memory = '32.GB'
    time   = '2.h'
  }
  withName: 'FASTTREE' {
    cpus   = 16
    memory = '16.GB'
  }
}
    

# Submit to SLURM cluster with Singularity
nextflow run your-github/BacTrack-NF \
  --input samplesheet.csv \
  --outdir /scratch/$USER/bactrack_out \
  -profile slurm,singularity \
  -resume

# Singularity images are pre-cached at:
# NXF_SINGULARITY_CACHEDIR=/path/to/cache

# Monitor run
nextflow log | tail -5
    

Developer

PhD researcher in
clinical metagenomics

I build reproducible bioinformatics tools at the intersection of clinical microbiology and computational genomics. BacTrack-NF grew out of frustration with fragmented tooling in a clinical WGS setting — I needed something I could hand to a lab technician with one command.

Open to remote bioinformatics positions — analyst, pipeline engineer, or computational researcher roles. Comfortable across the full stack: bash to Python to Nextflow to Docker.

Python R / Bioconductor Bash Nextflow DSL2 Snakemake Docker Singularity QIIME2 BLAST scikit-learn Git / GitHub Actions

6 Pipeline modules, each independently testable

~18 min Runtime for 12 samples on 32-core HPC node

4 Deployment profiles: local, Slurm, AWS, GCP

100% Processes containerised — no dependency conflicts

DSL2 Modern Nextflow with sub-workflows & nf-test

Clinical Bacterial WGSfrom raw reads to report

Clinical labs need a pipelinethat actually fits their workflow

Six modules, one command

Built for clinical reproducibility

Mock results from 12 K. pneumoniae isolates

Up and running in three commands

PhD researcher inclinical metagenomics

Tweaks

Clinical Bacterial WGS
from raw reads to report

Clinical labs need a pipeline
that actually fits their workflow

PhD researcher in
clinical metagenomics