RNA/DNA Variant Calling Pipeline

A modular SLURM-based pipeline for processing high-throughput sequencing data — Options including:

• QC and trimming raw FASTQ
• Alignment
• Mapping
• Feature quantification
• Variant calling (germline or somatic)

Built to support human and mouse data with runtime checks and SLURM job chaining.

---

Primarily this is for bespoke projects, not for regular/run of the mill pipelines run exactly the same way day in day out. Secondly this is used to help teach novice coders some of the basics, so it's designed to be slighly easier to break apart.

Feature	Bash	Nextflow	Snakemake
Simplicity and Learnability	✓ Simple, easy to teach.	✗ Requries learning JAVA-Groovy for Nextflow DSL.	✗ Requires learning Python for DSL workflow logic.
Flexibility and Control	✓ Full control over script.	✗ Abstracted control, less granular control.	✓ Flexible, configurable, some abstraction.
Resource Management	✓ Manual control of resources.	✓ Built-in resource management and scaling.	✓ Built-in resource management for scalability.
Debugging and Development	✓ Easier to debug errors.	✗ Can be harder to debug due to abstraction.	✓ Easier than Nextflow; more transparent errors.
Minimal Dependencies	✓ Only Singularity & SLURM.	✗ Requires installation of Nextflow.	✗ Requires installation of Snakemake.
Transparency and Portability	✓ Script is transparent.	✗ Harder to share due to Nextflow-specific configs.	✗ Harder to share due to Snakemake-specific configs.
Parallelism and Scalability	✗ Manual setup, not as scalable.	✓ Automatic parallelism and scalability.	✓ Automatic parallelism and scalability.
Community Support	✓ High levels of base code support.	✓ Growing support in bioinformatics.	✗ Mixed bioinformatics community with some workflows.
Reproducibility	✓ With use of Singularity.	✓ Reproducible with Singularity and environment management.	✓ Ensures reproducibility via dependency tracking.
Suitability for Bioinformatics	✓ Good for bespoke projects.	✓ Good for heavily used pipelines with dependencies.	✓ Useful for some specific bioinformatic tools.

---

Project Structure

project/
├── raw/                # Raw FASTQ input files
├── tmp/                # Temporary files
├── trim/               # Trimmed reads (fastp)
├── bam/                 # Aligned BAM files
├── counts/             # FeatureCounts output
├── vcf/                # DeepVariant/DeepSomatic VCFs
├── logs/               # Logs for SLURM and tools
└── SIFS/            # Main and helper scripts

---

Quick Start

0. Dependencies

• Singularity

Future development could include options for Local install, Conda and Docker.

1. Setup

git clone https://github.com/IainPerry/DNA_RNA_processing.git

#Optionally
cd SIF
chmod +x make_sifs.sh
./make_sifs.sh

Ensure modules or paths for Singularity, SLURM, and SIF containers are available. DEF files are available for container generation.

Update configuration variables in pipeline.sh.

# Run the pipeline interactively, in background
nohup bash pipeline.sh &
disown

It is recommended at first you try running interactively before disowning to check correct running.

2. Input Requirements

• FASTQ files: ${Sample}_M_F.fq.gz, ${Sample}_M_R.fq.gz
• Optional SOMATIC_TABLE.txt:

  tumor1 normal1
  tumor2 normal2
  

---

Pipeline Stages

Step	Tool	Output
Trimming + QC	fastp	Trimmed reads + HTML/JSON reports
Alignment	STAR or BWA	BAM files
Counting	featureCounts	Gene count tables
Variant Calling	DeepVariant / DeepSomatic	VCF files
Stats	bcftools stats	.stats files
Aggregation	MultiQC	Unified QC HTML report

---

Configurable Options

Variable	Description	Example
RUNTYPE	PE or SE	PE
SNPCALLING	Germline or Somatic	Somatic
MODEL	DeepVariant model type	WGS, WES
Species	Reference species	Human or Mouse
FASTArefH/M	Paths to human/mouse FASTA references	/path/ref.fa
Base, SIF_DIR	Mounted directories for Singularity
SOMATIC_TABLE	Path to matched tumor/normal table	./samples.txt

---

MultiQC Aggregates

The pipeline collects and aggregates the following into a final report:

• fastp: read trimming efficiency, quality distributions
• STAR / BWA: alignment summary
• featureCounts: assignment stats
• bcftools stats: variant calling summaries
• Optionally: Extend with FastQC, RSeQC, Picard

---

Example Output

logs/
├── trim/
│   ├── sample1_fastp.html
│   └── sample1_fastp.json
├── VCF/
│   ├── sample1.stats
│   └── slurm_sample1_index.out
multiqc/
└── multiqc_report.html

---

Contributions

Iain Perry

(Peter Giles - Some SIF recipies)

Additional planned stages include:

• Greater depth of QC metrics
• Dynamic outputting to webbased tools like Django

---

License

MIT License — free to use and adapt with attribution.

---