COMBINE-lab/salmon

What it solves

Salmon provides a fast and highly accurate way to quantify transcript abundances from RNA-seq data. It addresses the challenge of estimating how much of each transcript is present in a sample when dealing with raw sequencing reads or unsorted BAM alignments.

How it works

Salmon uses a two-stage process: first, it performs a fast mapping stage using either selective alignment or an alignment-free "sketch" mode. Second, it applies a massively-parallel statistical model (EM/VBEM over equivalence classes) to estimate the actual transcript abundances.

Who it’s for

It is designed for researchers and bioinformaticians working with RNA-seq data who need efficient, transcript-level quantification.

Highlights

• High Performance: Described as "wicked-fast" and rewritten in Rust for version 2.0.
• Flexible Input: Accepts raw sequencing reads or regular transcriptome alignments (unsorted BAM).
• Decoy-aware Indexing: Can index decoy sequences (like the genome) to prevent reads from being spuriously assigned to transcripts.
• Portable: Ships as a single portable binary with no external system library dependencies.