xgetSRAacc

#!/bin/bash
#

if [ $# -eq 0 ]; then
  echo ""
  echo "Usage: xgetSRAacc accession <'fasterq-dump options'>"
  echo ""
  echo "Examples:"
  echo "	xgetSRAcc SRR1270128"
  echo "		(will retrieve SRR1270128)"
  echo ""
  echo "	xgetSRAcc SRR1270128 '-t /dev/shm'"
  echo "		(will retrieve SRR1270128 using /dev/shm as tmp directory)"
  echo ""
  echo "	xgetSRAcc SRR1270128_SRR1270129_SRR1270130 '-e 8 -t /dev/shm'"
  echo "		(will retrieve SRR1270128, SRR1270129, and SRR1270130"
  echo "                 using 8 processors and then combine the data to files"
  echo "                 SRR1270128_SRR1270129_SRR1270130*.fastq using /dev/shm as"
  echo "                 tmp directory)"
  echo ""
  exit
fi 

SAMPLE=$1
FASTERQ_DUMP_OPTIONS=$2

echo ""
echo "Running xgetSRAcc ${SAMPLE} ${FASTERQ_DUMP_OPTIONS} ..."
echo ""

runs=${SAMPLE//_/ }

tmpdir=$(grep -o '\-t [^\S]*' <<< ${FASTERQ_DUMP_OPTIONS})
tmpdir=$(grep -o ' [^\S]*' <<< ${tmpdir})
if  [ -z "${tmpdir}" ]; then
  tmpdir=./
fi
echo "tmp directory set to:	${tmpdir}"
echo ""

for run in ${runs}; do
  echo "... attempting to get ${run} from NCBI SRA"
  n=1
  until [ $n -ge 8 ]; do
    echo ""
    echo "... attempt $n: if successful, ${run}*.fastq will have been deposited."
    echo ""
    fasterq-dump ${FASTERQ_DUMP_OPTIONS} ${run} && break
    touch ${tmpdir}/fasterq.tmp. && \rm -rf ${tmpdir}/fasterq.tmp.*
    n=$((n+1))
    sleep 30
  done
  if [ $n -eq 8 ]; then \
    echo "Time to figure out a different way to download the data from NCBI."; \
    echo "Or wait for better internet connection ..."; \
  else
    echo "... done"
  fi
done


runa=( ${runs} )
if [ ${#runa[@]} -gt 1 ]; then
  catcmd1="cat"
  catcmd2="cat"
  if [ -f ${run}_1.fastq ]; then
    for run in ${runs}; do
      catcmd1="${catcmd1} ${run}_1.fastq "
      catcmd2="${catcmd2} ${run}_2.fastq "
    done
    catcmd1="${catcmd1} > ${SAMPLE}_1.fastq"
    catcmd2="${catcmd2} > ${SAMPLE}_2.fastq"
    echo ""
    echo ${catcmd1}
    eval ${catcmd1}
    echo ${catcmd2}
    eval ${catcmd2}
  elif [ -f ${run}.fastq ]; then
    for run in ${runs}; do
      catcmd1="${catcmd1} ${run}.fastq"
    done
    catcmd1="${catcmd1} > ${SAMPLE}.fastq"
    echo ""
    echo ${catcmd1}
    eval ${catcmd1}
  else
    echo ""
    echo "Error: neither ${run}.fastq nor ${run}_1.fastq found."
    echo "Please check."
    echo ""
    exit
  fi
  if [ -f ${run}_1.fastq ]; then
    for run in ${runs}; do
      \rm ${run}_1.fastq ${run}_2.fastq
    done
  else
    for run in ${runs}; do
      \rm ${run}.fastq
    done
  fi
fi