Somatic Tumour/Normal Variant Calling (nf-core/sarek + Nextflow)

Matched tumour/normal (T/N) somatic variant calling workflow using nf-core/sarek, Nextflow, and Apptainer.

---

Repository structure

.
├── apptainer_cache/         # Apptainer images/cache
├── bin/                     # User-space shims (e.g., singularity -> apptainer)
├── conda_cache/             # Nextflow conda cache (if used)
├── metadata/                # Runinfo + samplesheets (CSVs)
├── nxf_work/                # Nextflow work directory
├── results/                 # Pipeline outputs (MultiQC, VCFs, pipeline_info)
├── scripts/                 # Execution & analysis scripts
└── README.md

---

Data

Study: SRP162370
Type: Paired-end FASTQs (matched T/N pair)

Sample	Accession	Role
Tumour	SRR8955957	Matched tumour
Normal	SRR8955958	Matched normal

Local FASTQs:

data/fastq_raw/
├── SRR8955957_1.fastq.gz   # tumour R1
├── SRR8955957_2.fastq.gz   # tumour R2
├── SRR8955958_1.fastq.gz   # normal R1
└── SRR8955958_2.fastq.gz   # normal R2

Download data (SRA → FASTQ)

This repo includes a helper script: scripts/Download_SRA.sh.

It:

• pulls RunInfo metadata
• downloads .sra files via prefetch
• converts to paired FASTQs via fasterq-dump
• compresses with pigz

---

Methodology

• Orchestration: Nextflow (local executor)
• Container execution: Apptainer via -profile singularity (shimmed if needed)
• Pipeline: nf-core/sarek
• Genome: GATK.GRCh38
• Callers: Mutect2, Strelka
• QC: MultiQC (FastQC, fastp, MarkDuplicates, etc.)

Key Sarek parameters:

--genome GATK.GRCh38
--tools mutect2,strelka

---

Usage

All scripts are in scripts/.

Start from the project root:

cd /mnt/vol1/WGS_variant_callling

1) Download data (SRA; FASTQ)

bash scripts/Download_SRA.sh

Outputs:

• metadata/SRP162370_runinfo.csv
• data/sra_raw/
• data/fastq_raw/

2) Test environment

Validates Nextflow + container execution end-to-end using a test profile.

bash scripts/run_sarek_test.sh

Outputs:

• results/sarek_test/

3) Generate samplesheet

Convert local FASTQs to Sarek CSV format.

bash scripts/02_make_samplesheets.sh

Outputs:

• metadata/samplesheet_tn_demo.csv
• metadata/samplesheet_normal_only.csv

4) Run pipeline (somatic calling)

Execute somatic calling on the tumour/normal pair.

bash scripts/02_run_sarek_somatic.sh

Outputs:

• results/tn_demo_somatic/

---

Results summary

QC metrics (MultiQC)

Note: This dataset is a low-coverage demonstration set (0X median), so treat results as workflow validation rather than a full biological interpretation.

Metric	Normal	Tumour
Reads	~3.4M	~4.0M
% mapped	~99.9%	~100.0%
Duplication (MarkDuplicates)	~95.4%	~95.6%
Median coverage	0X	0X

Variant counts

Callset	Total	PASS	SNV	INDEL
Mutect2 (filtered VCF)	1,044	444	697	347
Strelka (somatic SNVs)	2,160	497	2,160	0
Strelka (somatic indels)	18	1	0	18

Caller overlap (SNVs)

• Total overlap (all SNVs): 464
• PASS overlap (consensus subset): 228

---

Key outputs

Category	Path
QC report	results/tn_demo_somatic/multiqc/multiqc_report.html
Mutect2 VCF	results/tn_demo_somatic/variant_calling/mutect2/TUMOUR_vs_NORMAL/TUMOUR_vs_NORMAL.mutect2.filtered.vcf.gz
Strelka SNVs	results/tn_demo_somatic/variant_calling/strelka/TUMOUR_vs_NORMAL/TUMOUR_vs_NORMAL.strelka.somatic_snvs.vcf.gz
Strelka indels	results/tn_demo_somatic/variant_calling/strelka/TUMOUR_vs_NORMAL/TUMOUR_vs_NORMAL.strelka.somatic_indels.vcf.gz

---

Dependencies

Runtime

• Java 11+ (for Nextflow)
• Nextflow
• Apptainer

Data download

• SRA Toolkit (prefetch, fasterq-dump)
• Entrez Direct (esearch, efetch)
• pigz

Analysis

• R
• R Markdown

---

Scripts

• Download_SRA.sh — download SRP162370 metadata + tumour/normal reads, convert to FASTQs, compress
• run_sarek_test.sh — test profile run (environment validation)
• 02_make_samplesheets.sh — build Sarek samplesheets from local FASTQs
• 02_run_sarek_somatic.sh — run somatic calling (Mutect2 + Strelka)
• 03_tn_demo_somatic_report.Rmd — analysis + interpretation report

---