parallel-fastq-dump

parallel fastq-dump wrapper

Why & How

NCBI fastq-dump can be very slow sometimes, even if you have the resources (network, IO, CPU) to go faster, even if you already downloaded the sra file (see the protip below). This tool speeds up the process by dividing the work into multiple threads.

This is possible because fastq-dump have options (-N and -X) to query specific ranges of the sra file, this tool works by dividing the work into the requested number of threads, running multiple fastq-dump in parallel and concatenating the results back together, as if you had just executed a plain fastq-dump call.

Protips

• Downloading with fastq-dump is slow, even with multiple threads, it is recommended to use prefetch to download the target sra file before using fastq-dump, that way fastq-dump will only need to do the dumping.
• All extra arguments will be passed directly to fastq-dump, --gzip, --split-files and filters works as expected.
• This tool is not a replacement, you still need fastq-dump and sra-stat on your PATH for it to work properly.
• Speed improvements are better with bigger files, think at least 200k reads/pairs for each thread used.

Install

The preferred way to install is using Bioconda:

conda install parallel-fastq-dump

this will get you the sra-tools dependency as well.

Important: Make sure the sra-tools package being installed is a recent version (>=2.10.0) to guarantee compatibility with NCBI servers, conda might try to install an older version to be compatible with existing packages installed in your env, to be sure use this command:

conda install parallel-fastq-dump 'sra-tools>=3.0.0'

If that doesn't work you could also install it on a separate new env:

conda create -n testenv parallel-fastq-dump 'sra-tools>=3.0.0'

Examples

$ parallel-fastq-dump --sra-id SRR2244401 --threads 4 --outdir out/ --split-files --gzip

Micro Benchmark