Skip to content

A utility for easy downloading of reads from next-gen sequencing repositories like NCBI SRA

License

Notifications You must be signed in to change notification settings

louiejtaylor/grabseqs

Repository files navigation

grabseqs

Utility for simplifying bulk downloading data from next-generation sequencing repositories, like NCBI SRA, MG-RAST, and iMicrobe.

CircleCI Conda version Conda downloads Paper link

Note: read downloads for some samples from MG-RAST are not working through their web interface or API currently (3/20/2020)

Install

Install grabseqs and all dependencies via conda:

conda install grabseqs -c louiejtaylor -c bioconda -c conda-forge

Or with pip (and install the non-Python dependencies yourself):

pip install grabseqs

Note: If you're using SRA data, after you've installed sra-tools, run vdb-config -i and turn off local file caching unless you want extra copies of the downloaded sequences taking up space (read more here).

Quick start

Download all samples from a single SRA Project:

grabseqs sra SRP#######

Or any combination of projects (S/ERP), runs (S/ERR), BioProjects (PRJNA):

grabseqs sra SRR######## ERP####### PRJNA######## ERR########

If you'd like to do a dry run and just get a list of samples that will be downloaded, pass -l:

grabseqs sra -l SRP########

Similar syntax works for MG-RAST:

grabseqs mgrast mgp##### mgm#######

And iMicrobe (prefixing the sample numbers with "s" and project numbers with "p"):

grabseqs imicrobe p4 s3

Detailed usage

See the grabseqs FAQ for detailed troubleshooting tips!

Fun options:

grabseqs sra -t 10 -m metadata.csv -o proj/ -r 3 SRP#######

(translation: use 10 threads, save metadata to proj/metadata.csv, download to the dir proj/, retry failed downloads 3x, get all samples from SRP#######)

If you'd like to do a dry run and only get a list of samples that will be downloaded, pass -l:

grabseqs sra -l SRP########

If you'd like to pass your own arguments to fasterq-dump to get data in a slightly different format, you can do so like this:

grabseqs sra SRP####### -r 0 --custom_fqdump_args="--split-spot --progress"

Full usage:

grabseqs sra [-h] [-m METADATA] [-o OUTDIR] [-r RETRIES] [-t THREADS]
             [-f] [-l] [--no_parsing] [--parse_run_ids]
             [--use_fastq_dump]
             id [id ...]

positional arguments:
  id                One or more BioProject, ERR/SRR or ERP/SRP number(s)

optional arguments:
  -h, --help        show this help message and exit
  -m METADATA       filename in which to save SRA metadata (.csv format,
                    relative to OUTDIR)
  -o OUTDIR         directory in which to save output. created if it doesn't
                    exist
  -r RETRIES        number of times to retry download
  -t THREADS        threads to use (for fasterq-dump/pigz)
  -f                force re-download of files
  -l                list (but do not download) samples to be grabbed
  --parse_run_ids   parse SRR/ERR identifers (do not pass straight to fasterq-
                    dump)
  --custom_fqdump_args CUSTOM_FQD_ARGS
                    "string" containing args to pass to fastq-dump
  --use_fastq_dump  use legacy fastq-dump instead of fasterq-dump (no
                    multithreaded downloading)

Downloads .fastq.gz files to OUTDIR (or the working directory if not specified). If the -m flag is passed, saves metadata to OUTDIR with filename METADATA in csv format.

Similar options are available for downloading from MG-RAST:

grabseqs mgrast [-h] [-m METADATA] [-o OUTDIR] [-r RETRIES]
                [-t THREADS] [-f] [-l]
                rastid [rastid ...]

And iMicrobe:

grabseqs imicrobe [-h] [-m METADATA] [-o OUTDIR] [-r RETRIES]
                  [-t THREADS] [-f] [-l]
                  imicrobeid [imicrobeid ...]

Troubleshooting

See the grabseqs FAQ for detailed troubleshooting tips. If the FAQs don't fix your problem, feel free to open an issue!

Dependencies

  • Python 3 (external packages req'd: requests, requests-html, pandas, fake-useragent)
  • sra-tools>2.9
  • pigz
  • wget

If you use conda (on Linux), these will be installed for you!

Grabseqs runs on Mac or Linux. We've tested on these specific OSes:

Linux (conda or pip):

  • CentOS 6, 7, and 8
  • Debian 9 and 10
  • Ubuntu 16.04, 18.04, and 19.10
  • Red Hat Enterprise 6, 7, and 8
  • SUSE Enterprise 12 and 15

Mac (pip):

  • MacOS 10.14

Grabseqs has been tested and works with the following version of the Python dependencies (though these are neither minimal nor pinned version numbers):

  • requests 2.22.0
  • requests-html 0.10.0
  • pandas 0.25.3
  • fake-useragent 0.1.11

Citation

If you use grabseqs in your work, please cite:

Louis J Taylor, Arwa Abbas, Frederic D Bushman. "grabseqs: Simple downloading of reads and metadata from multiple next-generation sequencing data repositories." Bioinformatics, (2020), btaa167, https://doi.org/10.1093/bioinformatics/btaa167

Please also cite the researchers who generated the data (and the repository, if appropriate)!


Changelog

Dev version (not yet released)

  • Added a walk-through for adding a new repo using template.py
  • Better handling for invalid SRA accession numbers

0.7.0 (2020-01-29)

  • Allow users to pass custom args to fast(er)q-dump
  • Minor re-writes of download handling code for easier readability

0.6.1 (2019-12-20)

  • Validate compressed files (fix #8 and #34)

0.6.0 (2019-12-12)

  • Gracefully handle incomplete or missing dependencies
  • Major rewrite of test suite

0.5.2 (2019-12-05)

  • Improvements to work with multiple versions of Python 3

0.5.1 (2019-11-23)

  • Hotfix handling outdated versions of sra-tools

0.5.0 (2019-04-11)

  • Metadata available for all sources in .csv format

History

This project spawned out of/incorporates code from hisss; many thanks to ArwaAbbas for helping make this work!