Skip to content

🌶️ sracha 🌶️

Bioconda

Fast SRA downloader and FASTQ converter, written in pure Rust.

Features

  • Parallel downloads -- chunked HTTP Range requests with multiple connections
  • Native VDB parsing -- pure Rust, zero C dependencies
  • Integrated pipeline -- download, convert, and compress in one command
  • cSRA decode -- reference-compressed / aligned SRA archives decoded in pure Rust
  • ENA fast path -- --prefer-ena pulls pre-computed FASTQ.gz from ENA when available
  • Project-level accessions -- pass a BioProject (PRJNA) or study (SRP) to download all runs
  • Accession lists -- batch download from a file with --accession-list
  • Parallel gzip or zstd -- pigz-style block compression via rayon
  • FASTA output -- drop quality scores with --fasta
  • SRA and SRA-lite -- full quality or simplified quality scores
  • Split modes -- split-3, split-files, split-spot, interleaved
  • Stdout streaming -- pipe interleaved FASTQ to downstream tools with -Z
  • Resumable downloads -- automatically resumes interrupted transfers
  • File validation -- verify SRA file integrity with sracha validate
  • VDB introspection -- sracha vdb inspects .sra archives (tables, columns, metadata, schema) as a pure-Rust replacement for vdb-dump, either from a local file or straight from an accession over HTTP range requests, no download needed

How it works

sracha get runs the full pipeline in one command:

  1. Resolve -- looks up the accession via direct S3 URL (with SDL API fallback)
  2. Download -- fetches the .sra file with parallel chunked HTTP Range requests
  3. Parse -- reads the KAR archive and decodes VDB columns (READ, QUALITY, READ_LEN, NAME)
  4. Output -- formats FASTQ (or FASTA) records and compresses with parallel gzip/zstd

Demo

sracha get

Quick start

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
# Download, convert, and compress in one shot
sracha get SRR28588231

# Download all runs from a BioProject
sracha get PRJNA675068

# Batch download from an accession list
sracha get --accession-list SRR_Acc_List.txt

# Just download
sracha fetch SRR28588231

# Convert a local .sra file
sracha fastq SRR28588231.sra

# Show accession info
sracha info SRR28588231

# Validate a downloaded file
sracha validate SRR28588231.sra

See the Getting Started guide for more examples, or the CLI Reference for all options.

Installation

From binary releases

Download pre-built binaries from the releases page, which covers Linux (x86_64, aarch64), macOS (Intel, Apple silicon), and Windows (x86_64). Unix targets ship a .tar.gz; Windows ships a .zip holding a single statically linked sracha.exe with no runtime to install.

On Linux and macOS x86_64, pick the -v2 build unless you know the host has AVX2, in which case -v3 decodes faster. ARM and Windows builds ship a single binary.

From source

Requires Rust 1.95+.

1
cargo install --git https://github.com/rnabioco/sracha-rs sracha

With Bioconda

1
pixi add bioconda::sracha

Bioconda builds for Linux (x86_64, aarch64) and macOS (Intel, Apple silicon) only — it has no Windows channel, so this and the container images below are not available on Windows. Use the release .zip, cargo install, or WSL2.

With containers

Because sracha is on Bioconda, BioContainers automatically publishes a Docker/Singularity image for every release — no local build required.

1
2
3
4
5
6
# Docker / Podman
docker run --rm quay.io/biocontainers/sracha:0.7.0--h54198d6_0 sracha --help

# Singularity / Apptainer
singularity run \
  https://depot.galaxyproject.org/singularity/sracha:0.7.0--h54198d6_0 sracha --help

These tags track the current Bioconda release and are refreshed weekly by a CI job. BioContainers builds an image a day or two behind a new release, so just after a release the tag here may still name the previous version — see quay.io for every published <version>--<build> tag.

Nextflow

In Nextflow, point a process at the image directly or let the conda directive resolve it:

1
2
3
4
5
process SRACHA_GET {
    container 'quay.io/biocontainers/sracha:0.7.0--h54198d6_0'
    // or: conda 'bioconda::sracha=0.7.0'
    // ...
}

Pin the version for reproducibility, as above. Dropping the version from the conda directive (conda 'bioconda::sracha') resolves to whatever is current in Bioconda instead, which is convenient for ad-hoc runs but makes the pipeline non-reproducible.

Acknowledgments

sracha builds on the Sequence Read Archive, maintained by the National Center for Biotechnology Information at the National Library of Medicine. The SRA and its toolchain are public-domain software developed by U.S. government employees — our tax dollars at work. Special thanks to Kenneth Durbrow (@durbrow) and the SRA Toolkit team for building and maintaining the infrastructure that makes projects like this possible.

This project wouldn't exist without NCBI's open infrastructure: the VDB/KAR format, the SDL locate API, EUtils, and public S3 hosting of sequencing data. sracha aims to make it easier for the community to build on that foundation.