parallel fastq-dump wrapper
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.
- Downloading with
fastq-dumpis slow, even with multiple threads, it is recommended to useprefetchto download the target sra file before usingfastq-dump, that wayfastq-dumpwill only need to do the dumping. - All extra arguments will be passed directly to
fastq-dump,--gzip,--split-filesand filters works as expected. - This tool is not a replacement, you still need
fastq-dumpandsra-staton yourPATHfor it to work properly. - Speed improvements are better with bigger files, think at least 200k reads/pairs for each thread used.
The preferred way to install is using Bioconda:
conda install parallel-fastq-dump
this will get you the sra-tools dependency as well.
$ parallel-fastq-dump --sra-id SRR2244401 --threads 4 --outdir out/ --split-files --gzip
