Download FASTQ files from the EBI SRA mirror

get_sra.sh

#!/usr/bin/env bash

get_sra() {
  local __FASTQ="$1"

  local __OUT="$PWD"
  local __DOWNLOAD_THREADS=1

  # Allow downloading with multiple threads
  if [[ ! -z "${DOWNLOAD_THREADS+x}" ]]; then
    __DOWNLOAD_THREADS=${DOWNLOAD_THREADS}
  fi

  # Allow optional output dir
  if [[ $# -eq 2 ]]; then
    __OUT="$2"
  fi

  # Figure out the location on the EBI server
  local __PREFIX="ftp://ftp.sra.ebi.ac.uk/vol1/fastq"
  local __ACCESSION=""
  __ACCESSION=$(echo "${__FASTQ}" | tr '.' '_' | cut -d '_' -f 1)
  local __DIR1="${__ACCESSION:0:6}"
  local __A_LEN=${#__ACCESSION}
  local __DIR2=""
  if [[ ${__A_LEN} -eq 9 ]]; then
    __DIR2=""
  elif [[ ${__A_LEN} -eq 10 ]]; then
    __DIR2="00${__ACCESSION:9:1}"
  elif [[ ${__A_LEN} -eq 11 ]]; then
    __DIR2="0${__ACCESSION:9:2}"
  else
    __DIR2="${__ACCESSION:9:3}"
  fi
  local __URL="${__PREFIX}/${__DIR1}/${__DIR2}/${__ACCESSION}/${__FASTQ}"

  # Check if we are PE or SE
  local __STATUS
  __STATUS=$(wget --spider "${__URL}_2.fastq.gz" 2>&1 | grep exists || true)

  # Set appropriate variables and download
  if [[ ! -z "${__STATUS}" ]]; then
    PAIRED=1
    FIRST="${__OUT}/${__FASTQ}_1.fastq.gz"
    SECOND="${__OUT}/${__FASTQ}_2.fastq.gz"
    printf "%s\n%s\n" "${__URL}_1.fastq.gz" "${__URL}_2.fastq.gz" |
      xargs -I {} -P 2 wget --quiet -P "${__OUT}" {}
  else
    PAIRED=0
    FIRST="${OUT}/${FASTQ}.fastq.gz"
    SECOND=""
    wget --quiet -P "${OUT}" "${URL}.fastq.gz"
  fi
}