Salmon

Overview

Salmon is a fast and bias-aware RNA-seq quantification tool that estimates transcript-level abundances using selective alignment (mapping-based mode) or quasi-mapping. It accounts for common biases in RNA-seq data including fragment GC content bias, positional bias, and sequence-specific bias through its built-in correction models. Salmon can also operate in alignment-based mode, taking a pre-aligned BAM file as input. Its speed and accuracy make it a standard choice in RNA-seq pipelines, and its output integrates directly with tximport for gene-level summarisation and downstream differential expression analysis with DESeq2 or edgeR.

Installation

mamba install -c bioconda salmon

Basic Usage

Build a transcriptome index

salmon index -t transcriptome.fa -i salmon_index -p 8

Quantify paired-end reads with selective alignment

salmon quant -i salmon_index -l A \
  -1 sample_R1.fastq.gz -2 sample_R2.fastq.gz \
  -p 8 --validateMappings \
  -o salmon_output/

The -l A flag enables automatic library type detection.

Key Parameters

Flag / option	Description
`index -t`	Path to the transcriptome FASTA for building the index.
`index -i`	Output path for the Salmon index directory.
`quant -i`	Path to the pre-built Salmon index.
`-l`	Library type (`A` for automatic detection, or explicit types such as `ISR`, `ISF`, `IU`).
`-1` / `-2`	Paired-end read files (forward and reverse).
`-r`	Single-end read file.
`-p`	Number of threads to use.
`--validateMappings`	Enable selective alignment for improved mapping accuracy.
`-o`	Output directory for quantification results.
`--gcBias`	Enable GC bias correction (recommended for most datasets).
`--seqBias`	Enable sequence-specific bias correction.
`--numBootstraps`	Number of bootstrap samples for variance estimation.

Expected Output

Salmon writes the following files to the output directory:

quant.sf – the primary output file: a tab-delimited table with columns for transcript name, length, effective length, TPM (transcripts per million), and estimated read counts (NumReads).
quant.genes.sf – gene-level quantification (when a gene map is provided).
aux_info/ – directory containing auxiliary information including the observed library type, fragment length distribution, bias correction parameters, and the equivalence class file.
cmd_info.json – a JSON file recording the exact command and parameters used for the run.
logs/ – directory containing log files with mapping rate and run statistics.