nf-core pipeline · rna, rna-seq
nf-core/rnaseq
A bioinformatics pipeline for quality control, alignment, and quantification of reference-based RNA sequencing data.
What nf-core/rnaseq is
nf-core/rnaseq is a Nextflow bioinformatics pipeline engineered to process bulk RNA sequencing data from organisms with an existing reference genome and annotation. It accepts FASTQ files or pre-aligned BAM files and performs quality control, contaminant removal, alignment or pseudoalignment, and feature quantification.
The pipeline automates complex multi-stage workflows required to transform raw sequence reads into normalized gene and isoform expression matrices. In addition to expression quantification, it performs extensive quality control across raw sequence reads, alignments, biotype allocations, and sample similarities, gathering diagnostics into comprehensive MultiQC reports. It prepares standardized count data ready for downstream statistical and differential expression tools.
Designed for high reproducibility and flexible execution across environments, the workflow supports multiple alignment routes (such as STAR and HISAT2) and pseudoalignment alternatives (Salmon and Kallisto). It standardises processing, reduces manual intervention, and resolves technical hurdles like automatic strandedness inference and technical replicate merging.
What the pipeline does
The workflow begins by merging any technical replicates listed under identical sample names in a CSV samplesheet, followed by automated strandedness inference using fq and Salmon. Raw read quality is evaluated with FastQC, unique molecular identifiers (UMIs) are extracted if present using UMI-tools, and adapters and low-quality bases are trimmed with Trim Galore!. Contaminants and ribosomal RNA can optionally be identified and removed via BBSplit and SortMeRNA before sequence mapping.
Reads are then processed through user-selected alignment or pseudoalignment pipelines. Alignment pathways include STAR paired with quantification via Salmon or RSEM, or HISAT2 for alignment only; alternatively, direct transcript-level pseudoalignment is supported via Salmon or Kallisto. Following genomic alignment, SAMtools handles coordinate sorting and indexing, Picard MarkDuplicates or UMI-tools marks or removes duplicates, StringTie performs transcript assembly and quantification, and BEDTools produces bigWig coverage tracks. Broad quality metrics are collected across RSeQC, Qualimap, dupRadar, Preseq, DESeq2 (or alternatively the single-pass experimental tool RustQC), and screening for foreign organisms via Kraken2/Bracken or Sylph, with all summaries compiled into a MultiQC report.
Key facts
- Pipeline framework
- Nextflow DSL2
- Input formats
- FASTQ (single-end or paired-end), BAM (pipeline-generated pre-aligned)
- Reference inputs
- FASTA, GTF
- Primary outputs
- Gene expression matrix, isoform counts, bigWig coverage files, MultiQC report
- Aligners
- STAR, HISAT2
- Quantifiers
- Salmon, RSEM, StringTie, Kallisto
- Latest release
- 3.26.0 (2026-05-07)
- Maintainer
- Jonathan Manning and the nf-core community
What you can use it for
- Reference-guided gene and transcript quantification
- Processing raw paired-end or single-end RNA-seq FASTQ reads against an annotated reference genome to derive gene- and isoform-level count matrices using STAR coupled with Salmon or RSEM.
- Fast lightweight pseudoalignment
- Quantifying transcript abundance directly from raw reads using pseudoaligners like Salmon or Kallisto, bypassing genomic coordinate alignment.
- Two-step re-analysis from pipeline BAM files
- Reprocessing data using pre-aligned BAM files generated in prior runs with intermediate saving enabled, skipping alignment to run downstream QC and assembly tools quickly.
- Contamination and metagenomic screening
- Detecting sample contaminants or microbial sequences from screening reads using Kraken2 paired with Bracken, or Sylph.
Run it here
The research agent runs nf-core/rnaseq with Nextflow on its own machine — up to 32 cores, 256 GiB of memory and 24 hours per run, which covers the test profile and real single-sample runs. It builds the samplesheet, checks it against the pipeline's schema and reads the results back. When a run is too big for one machine, it gives you the exact command for your own cluster instead.
Strengths and limits
Good at
- Supports multiple aligner and quantifier combinations including STAR, RSEM, HISAT2, Salmon, and Kallisto.
- Automatically infers library strandedness by subsampling reads and pseudoaligning them with Salmon.
- Merges technical replicates automatically when provided with shared sample IDs in the samplesheet.
- Offers comprehensive quality control evaluation through RSeQC, Qualimap, dupRadar, DESeq2, and MultiQC.
- Handles unique molecular identifier (UMI) extraction and deduplication using UMI-tools.
Watch out for
- Does not calculate gene or transcript expression estimates when running the HISAT2 alignment route.
- Requires an organism with an existing reference genome and GTF annotation file.
- Does not perform comparative statistical hypothesis testing, such as calculating differential expression P-values or false discovery rates (FDR).
- Pre-aligned BAM reprocessing requires intermediate BAM files created specifically by this pipeline.
Questions
Can I run nf-core/rnaseq without my own cluster?
The research agent runs nf-core/rnaseq with Nextflow on its own machine — up to 32 cores, 256 GiB of memory and 24 hours per run, which covers the test profile and real single-sample runs. It builds the samplesheet, checks it against the pipeline's schema and reads the results back. When a run is too big for one machine, it gives you the exact command for your own cluster instead.
What licence is nf-core/rnaseq under?
MIT, like every nf-core pipeline.
Can this pipeline perform differential expression analysis between conditions?
No. The pipeline outputs gene and transcript count matrices and normalises read data, but it does not run comparative statistical tests to calculate P-values or false discovery rates. Downstream statistical analysis can be performed in environments like R or Julia, or by feeding the outputs into nf-core/differentialabundance.
Why is there no expression quantification produced when using HISAT2?
Quantification is disabled for the HISAT2 route due to the absence of an appropriate method to compute accurate expression estimates directly from HISAT2-derived genomic alignments. It is intended for alignment, quality control, and downstream analyses suited specifically to HISAT2 output.
How does the pipeline handle library strandedness?
When the strandedness field in the samplesheet is set to 'auto', the pipeline automatically infers the strand configuration using a subsampling pseudoalignment procedure via fq and Salmon.
Can existing BAM files be provided directly as pipeline input?
Yes, but this two-step workflow is designed specifically for BAM files previously generated by nf-core/rnaseq using the intermediate save flag, allowing users to skip alignment and run downstream analyses without repeated compute costs.
Can public SRA accessions be downloaded directly within nf-core/rnaseq?
No. Direct SRA download functionality was moved to the separate nf-core/fetchngs workflow, which can automatically produce a compatible samplesheet for nf-core/rnaseq.
How to cite it
10.5281/zenodo.1400710
Written from nf-core/rnaseq's own README. Statements citing a figure that does not appear there are removed before publication. Licence and running details come from our registry.