sc2-sequencing/ont/ont-transfer

#!/bin/bash
# Copies fast5 and fastq files from a GridION to the first available USB device

# Find the skipped barcodes parameter
if [[ ${1} = "-s" ]]; then
  OFFSET=${2}
  FOLDERIN=${3}
else
  echo " Assuming no skipped barcodes"
  echo "   CTRL+C if incorrect and call 'ont-transfer -s [1|6|12]', instead"
  OFFSET=1
  FOLDERIN=${1}
fi

# Find the directory we are supposed to copy
# If no directory was passed, check to see if the current directory will work
if [[ ! -n "$FOLDERIN" ]]; then
  if [[ -n $(find . -type d -name "fast5_pass") ]]; then
    FOLDERPATH="$PWD"
    FOLDERNAME=$(basename "$PWD")
  else
    echo " ERROR: Can't find FAST5s in current directory and no directory was passed."
    echo "   ont-transfer must either be passed a directory or be called from a directory"
    echo "   containing a 'fast5_pass' folder containing FAST5 files."
    exit 1
  fi
else
  FOLDERPATH="$FOLDERIN"
  FOLDERNAME=$(basename "$FOLDERIN")
fi

# Calculate the FASTAs that need copied
if [[ $OFFSET -gt 1 ]]; then
  if [[ $OFFSET -lt 7 ]]; then
    KEEPERS=(01 02 03 04 05 06)
  else
    KEEPERS=(07 08 09 10 11 12)
  fi
  for ((i = 1; i < 8; ++i)); do
    WORKING=(${KEEPERS[@]: -6})
    for ((j = 0; j < 6; ++j)); do
      KEEPERS+=($(( 10#${WORKING[j]} + 12 )))
    done
  done
else
  KEEPERS=( $(seq -f "%02g" 01 96) )
fi

# Find where the flash drive is mounted
# Note that this only works on GridIONs thanks to goofy mounting
USBDRIVE=$(mount | grep /data | awk '{print $3}' | tail -n1)
if [[ -n $(echo $USBDRIVE | grep scratch) ]]; then
  echo "No USB drive was detected. Exiting now."
  exit 1
fi

# Prompt the users
echo "Extracting FAST5/FASTQ from $FOLDERNAME to USB Device $USBDRIVE, skipping after $OFFSET."
echo "If this is not correct, press CTRL+C within the next 10 seconds to abort..."
sleep 10
echo "Proceeding..."

# Simplify the output paths
USBPATH="$USBDRIVE/$FOLDERNAME"
FAST5PATH="$USBPATH/fast5"
FASTQPATH="$USBPATH/fastq"

# Make output directories
mkdir -p "$FAST5PATH"
mkdir -p "$FASTQPATH"

# Copy the files
for FASTA in ${KEEPERS[@]}; do
  # Check to see if there are any files here
  if [[ -n $(find $FOLDERPATH -type f -name "*.fast5" -path "*_pass*barcode$FASTA*") ]]; then
    # Find the uid that the GridION gives to this sample
    ONTID=$(find $FOLDERPATH -name "*.fast5" -path "*_pass*barcode*$FASTA*" | head -n 1 | xargs basename | awk '{split($0,a,"_"); print a[4]}')

    # Copy FAST5s
    find $FOLDERPATH -name "*.fast5" -path "*_pass*barcode$FASTA*" -exec cp -n {} $FAST5PATH \;

    # Unzip any gzipped fastqs
    find $FOLDERPATH -name "*.fastq.gz" -path "*_pass*barcode$FASTA*" -print0 | xargs -0 -r -L1 -P0 gunzip

    # Get the name of the resulting FASTQ file
    FASTQRESULT="$FASTQPATH"/"$FOLDERNAME"_pass_barcode"$FASTA"_"$ONTID"_0.fastq

    # Concatenate the fastqs
    cat $FOLDERPATH/*/fastq_pass/barcode$FASTA/*.fastq > $FASTQRESULT

    # Zip up the remainders
    find $FOLDERPATH -name "*.fastq" -path "*_pass*barcode$FASTA*" -print0 | xargs -0 -r -L1 -P0 gzip

    # Zip up the copied fastq
    gzip -f $FASTQRESULT
  fi
done