SPLIT 🧬

SNP-Level Inspection of Parental Transcripts

Nextflow DSL2 pipeline for allele-specific RNA-seq analysis using STAR, SNPsplit, and featureCounts. Ultra-minimalist — 2 files only (main.nf + nextflow.config). Designed for solo bioinformaticians.

---

Pipeline Overview

%%{init: {'theme': 'base', 'themeVariables': {'background': '#ffffff', 'primaryTextColor': '#231f20', 'lineColor': '#999999', 'textColor': '#231f20', 'mainBkg': '#ffffff', 'nodeBorder': '#999999'}}}%%
flowchart TD
    subgraph INPUT ["Input (one of)"]
        SRA["SRR / ERR / DRR"] --> SRA_DL["SRA_DOWNLOAD"]
        GEO["GSE / GSM"] --> RESOLVE["RESOLVE_GEO"] --> SRA_DL
        FQ_DIR["FASTQ directory"]
        CSV["CSV samplesheet"]
    end

    SRA_DL --> FASTQS(("FASTQs"))
    FQ_DIR --> FASTQS
    CSV --> FASTQS

    DOWNLOAD["DOWNLOAD_REFERENCES"] --> GPREP["SNPSPLIT_GENOME_PREP"]
    GPREP --> IDX1["STAR_INDEX (N-masked)"]
    DOWNLOAD --> IDX2["STAR_INDEX (reference)"]

    IDX1 --> A1["STAR_ALIGN (N-masked)"]
    IDX2 --> A2["STAR_ALIGN (reference)"]
    FASTQS --> A1
    FASTQS --> A2

    A1 --> S1["SORT_DEDUP"] --> SNP["SNPSPLIT"]
    A2 --> S2["SORT_DEDUP "] --> FC3["FEATURECOUNTS (reference)"]

    SNP -->|"genome1"| FC1["FEATURECOUNTS (genome1)"]
    SNP -->|"genome2"| FC2["FEATURECOUNTS (genome2)"]

    FC1 --> O1["genome1 counts"]
    FC2 --> O2["genome2 counts"]
    FC3 --> O3["reference counts"]
    FC1 & FC2 & FC3 --> MQC["MULTIQC"] --> O4["MultiQC report"]

    classDef input fill:#0570b0,stroke:#0570b0,color:#fff
    classDef process fill:#238b45,stroke:#238b45,color:#fff
    classDef key fill:#cb181d,stroke:#cb181d,color:#fff,stroke-width:3px
    classDef output fill:#6a51a3,stroke:#6a51a3,color:#fff
    classDef data fill:#e6550d,stroke:#e6550d,color:#fff
    classDef mqc fill:#41ab5d,stroke:#41ab5d,color:#fff

    class SRA,GEO,FQ_DIR,CSV input
    class SRA_DL,RESOLVE,DOWNLOAD,GPREP,IDX1,IDX2,A1,S1,A2,S2,FC1,FC2,FC3 process
    class SNP key
    class O1,O2,O3,O4 output
    class FASTQS data
    class MQC mqc

Loading

Quick Start

# From a FASTQ directory (auto-detects PE/SE)
nextflow run IPNP-BIPN/SPLIT --fastq_dir /path/to/fastqs --outdir results -resume

# From SRA accessions
nextflow run IPNP-BIPN/SPLIT --sra_ids "SRR1234567,SRR1234568" --outdir results -resume

# From a GEO dataset (auto-resolves GSE → SRR)
nextflow run IPNP-BIPN/SPLIT --sra_ids GSE80810 --outdir results -resume

# From a samplesheet CSV
nextflow run IPNP-BIPN/SPLIT --input samplesheet.csv --outdir results -resume

# Custom strains
nextflow run IPNP-BIPN/SPLIT \
    --fastq_dir /path/to/fastqs \
    --strain1 CAST_EiJ \
    --strain2 C57BL_6NJ \
    --dedup true \
    --outdir results \
    -resume

Samplesheet format (CSV)

sample,fastq_1,fastq_2
sampleA,/path/to/sampleA_R1_001.fastq.gz,/path/to/sampleA_R2_001.fastq.gz
sampleB,/path/to/sampleB.fastq.gz,

Leave fastq_2 empty for single-end reads.

---

Parameters

Parameter	Default	Description
--input	null	Samplesheet CSV (sample,fastq_1,fastq_2)
--fastq_dir	null	Directory of FASTQs (*.fastq.gz)
--sra_ids	null	SRA/GEO accessions (comma-separated or file)
--outdir	results	Output directory
--strain1	CAST_EiJ	First strain in VCF → genome1
--strain2	C57BL_6NJ	Second strain in VCF → genome2
--dedup	true	Remove PCR duplicates (samtools markdup -r)
--strandedness	0	featureCounts strandedness (0/1/2)
--force_se	false	Force single-end counting in featureCounts
--genome_url	Ensembl GRCm39	Genome FASTA URL
--gtf_url	Ensembl 2023_04	GTF annotation URL
--vcf_url	MGP REL2021 v8	VCF SNPs URL
--star_limit_genome_ram	60000000000	STAR --limitGenomeGenerateRAM
--max_cpus	auto	Maximum number of CPUs
--max_memory	64 GB	Maximum memory (scales all process labels)

Pre-built references (skip downloads)

Parameter	Description
--genome_fa	Pre-downloaded genome FASTA
--gtf	Pre-downloaded GTF
--vcf	Pre-downloaded VCF
--star_index_nmask	Pre-built STAR index (N-masked)
--star_index_ref	Pre-built STAR index (reference)
--snp_file	Pre-existing SNPsplit SNP annotation file

---

Output Structure

results/
├── 00_sra_fastq/           # Downloaded FASTQs (if SRA input)
├── 04_aln_nmask/           # N-mask aligned BAMs (sorted, deduped)
├── 05_aln_ref/             # Reference aligned BAMs (sorted, deduped)
├── 06_snpsplit/            # SNPsplit output (genome1, genome2, unassigned)
├── 07_counts/
│   ├── counts_genome1_CAST_EiJ.txt      # Allele-specific counts (strain1)
│   ├── counts_genome2_C57BL_6NJ.txt     # Allele-specific counts (strain2)
│   └── counts_reference.txt              # Standard reference counts
├── 08_multiqc/             # Aggregated QC report
├── reference/              # Downloaded + cached references
│   ├── genome.fa           # GRCm39 soft-masked
│   ├── genes.gtf           # Ensembl annotation
│   ├── snps.vcf.gz         # MGP REL2021 SNPs
│   ├── snpsplit_prep/      # N-masked genome + SNP file
│   ├── star_nmask/         # STAR index (N-masked)
│   └── star_ref/           # STAR index (reference)
└── pipeline_info/          # Nextflow timeline, trace, DAG, report

---

Requirements

Core (always required): STAR samtools SNPsplit featureCounts (subread) multiqc wget

Optional: sra-tools bgzip (htslib/tabix) — for SRA download

Nextflow ≥ 23.04

---

How it works

1. References are automatically downloaded from Ensembl (GRCm39 genome + GTF) and MGP (SNP VCF). Cached via storeDir — only downloaded once.

2. SNPsplit genome preparation creates an N-masked genome where strain-discriminating SNP positions are replaced by N. This prevents alignment bias toward the reference allele.

3. Two parallel alignment tracks:

   • N-masked track: alignments used for allele-specific analysis (SNPsplit)
   • Reference track: standard alignments for total gene expression

4. SNPsplit assigns each read from the N-mask track to genome1 (strain1), genome2 (strain2), or unassigned based on informative SNP positions.

5. featureCounts produces three count tables using gene_name attribute for human-readable gene symbols (e.g., Gapdh instead of ENSMUSG00000057666).

---

Resume & Cache

The pipeline natively leverages Nextflow's cache (-resume). Already completed steps are automatically skipped. References (genome, GTF, VCF, STAR indexes, N-masked genome) are persisted via storeDir and reused across runs.

# Re-run after a crash — picks up exactly where it left off
nextflow run main.nf --fastq_dir fastqs --outdir results -resume

---

License

MIT