RNA-Seq Neoantigen Discovery Pipeline

End-to-end Nextflow DSL2 pipeline for somatic variant calling and MHC-I neoantigen prediction from murine RNA-Seq data (GRCm39 / Mus musculus). Designed for personalised cancer immunotherapy research.

---

Pipeline Overview

FastQ → FastQC + Trimmomatic → STAR (or NVIDIA Parabricks)
     → Strelka2 (Somatic/Germline) → GATK Hard-Filtering
     → VEP Annotation → Mutant Peptide Extraction
     → NetMHCpan-4.1 Scoring → Ranked Neoantigen Candidates

Stage	Tool	Output
Quality Control	FastQC, Trimmomatic	QC reports, trimmed FASTQ
Alignment	STAR 2.7.11 / NVIDIA Parabricks rna_fq2bam	Sorted, indexed BAM
Variant Calling	Strelka2 2.9.10	Somatic + germline VCF
Variant Filtering	GATK 4.5 VariantFiltration	PASS-only VCF
Annotation	Ensembl VEP 111	Functionally annotated VCF
Peptide Extraction	Ensembl REST API + vcf_to_peptides.py	Mutant peptide FASTA
Neoantigen Scoring	NetMHCpan-4.1 + run_netmhcpan.py	Ranked MHC-I candidate TSVs

---

Requirements

• Nextflow >= 23.04.0
• Docker or Singularity
• Java >= 17
• NetMHCpan-4.1 — academic licence required (see Docker setup)
• (Optional) NVIDIA GPU + Clara Parabricks licence for GPU-accelerated alignment

---

Quick Start

# 1. Clone
git clone https://github.com/pOSNode/Mouse_Neoantigen_Candidates.git
cd Mouse_Neoantigen_Candidates

# 2. Run test profile (uses bundled chr19 test data)
nextflow run main.nf -profile test,docker

# 3. Full run
nextflow run main.nf \
    -profile docker \
    --reads  '/data/fastq/*_R{1,2}.fastq.gz' \
    --outdir results

---

Parameters

Parameter	Default	Description
--reads	null	Glob pattern for paired FASTQ files
--outdir	results	Output directory
--genome	—	Reference FASTA (GRCm39)
--star_index	—	Pre-built STAR index directory
--use_parabricks	false	Use GPU-accelerated alignment
--qual_threshold	30	GATK QUAL hard-filter threshold
--dp_threshold	10	GATK DP hard-filter threshold
--mhc_alleles	H-2-Kb,H-2-Db	Murine MHC-I alleles
--peptide_lengths	8,9,10,11	MHC-I peptide lengths
--peptide_flank	13	AA flank around mutation site
--el_rank_cutoff	2.0	%EL Rank filter (<=0.5 strong binder)
--ba_rank_cutoff	2.0	%BA Rank filter (~500 nM IC50)

---

Profiles

Profile	Description
docker	Run with Docker containers
singularity	Run with Singularity containers (HPC)
gcp	Google Cloud Batch executor
hpc	SLURM executor
test	Local test run with bundled chr19 data

---

Docker Setup

All containers except NetMHCpan pull automatically from public registries when running with -profile docker or -profile singularity.

NetMHCpan-4.1 — manual build required

Licence notice: NetMHCpan-4.1 is free for academic, non-commercial use only. The binary is not redistributable and therefore not included in this repository and not hosted on any public container registry. You must obtain your own copy directly from DTU Health Tech and build the image locally before running the neoantigen scoring step.

# 1. Register and download the Linux binary (free for academic use):
#    https://services.healthtech.dtu.dk/services/NetMHCpan-4.1/
#    Place the downloaded tarball at: docker/netMHCpan-4.1b.Linux.tar.gz

# 2. Build the image locally — this must be done before running the pipeline
docker build -f docker/Dockerfile.netmhcpan -t owais/netmhcpan:4.1 .

# 3. Verify
docker run --rm owais/netmhcpan:4.1 netMHCpan -version

If you attempt to run the pipeline without building this image first, the NETMHCPAN process will fail with a container-not-found error.

---

Outputs

results/
├── fastqc/               # Per-sample FastQC HTML + ZIP reports
├── trimmomatic/          # Trimmed FASTQs + trimming logs
├── star/                 # Aligned BAMs + STAR logs
├── strelka2/             # Somatic + germline VCFs
├── gatk/                 # PASS-filtered VCFs
├── vep/                  # VEP-annotated VCFs + HTML summaries
├── netmhcpan/
│   ├── *_netmhcpan_raw.tsv             # All predictions (pre-filter)
│   ├── *_netmhcpan_candidates.tsv      # PASS candidates, ranked by priority score
│   └── *_netmhcpan_summary.tsv         # Per-gene/allele best-binder summary
└── pipeline_info/        # Nextflow timeline, report, trace, DAG

Candidate Output Columns

Column	Description
rank	Priority rank across all allele/length combinations
gene	Gene symbol (e.g. Hsp90ab1, Ncl)
allele	MHC-I allele (e.g. H-2-Kb)
peptide_length	Mer length (8-11)
peptide	Presented peptide sequence
core	NetMHCpan binding core (9-mer)
el_rank	%EL Rank — predicted ligand likelihood (<=0.5 strong binder)
ba_rank	%BA Rank — predicted binding affinity rank (<=2.0 ~500 nM)
priority_score	Composite score: 0.6x(1-EL/100) + 0.3x(1-BA/100) + 0.1x(1/len)
hgvsp	HGVSp notation of the causal variant
consequence	VEP consequence term
bind_level	NetMHCpan classification: SB strong / WB weak binder

---

Citation

If you use this pipeline, please cite:

Siddiqi O. (2024). RNA-Seq Neoantigen Discovery Pipeline (v1.0.0). GitHub. https://github.com/pOSNode/Mouse_Neoantigen_Candidates ORCID: 0009-0006-4754-8200

---

Licence

MIT © Owais Siddiqi