Create an adapter sequence file from the barcodes in sample names

barcodeToAdapter.sh

#!/bin/bash

set -e
set -o pipefail

###########################################################################################################################
#########################################           Usage Information             #########################################
#   This is a script to extract adapter sequences that use barcodes embedded in the names of files
#   To use, add a list of samples to the SAMPLE_INFO field on line 34
#       This should look like:
#           SAMPLE_INFO=${HOME}/directory/sample_list.txt
#   Set the smallest size barcode in the MIN_BAR field on line 37
#       This shoudl look like:
#           MIN_BAR=6
#   Define an outfile in the OUT field on line 41
#       This should look like:
#           OUT=${HOME}/directory/out_adapters.txt
#       This defaults to a file in the current working directory called extracted_adapters with the date and time attached
#   If there's a sequence before the barcode in the adapter sequence, type it into the PRECEDING_SEQ field on line 45
#       This should look like:
#           PRECEDING_SEQ="ACTG"
#   If there's a sequence trailing the barcode in the adapter sequence, type it into the FOLLOWING_SEQ field on line 49
#       This should look like:
#           FOLLOWING_SEQ="ACTG"
#   To set a base name for the adapter sequences, type it into the HEADER field on line 53
#       This should look like:
#           HEADER=Adapter
#       Each adapter sequence found will be called ${HEADER}(sequential numbers from 0 to the number of sequences)
#   To run, type the following command:
#       bash barcodeToAdapter.sh
###########################################################################################################################

#   Enter a sample list
SAMPLE_INFO=

#   Define the smallest size barcode
MIN_BAR=2

#   Set an output file
#       This is optional
OUT=

#   Set a sequence that comes immediately before the barcode in the adapter sequence
#       This is optional
PRECEDING_SEQ=

#   Set a sequence that comes immediately after the barcode in the adapter sequence
#       This is optional
FOLLOWING_SEQ=

#   Set a title to call each adapter sequence
#       This is optional
HEADER=

###########################################################################################################################
####################                    Do the work here                                               ####################
####################                    Don't change anything here                                     ####################
####################                    Unless you know what you're doing                              ####################
###########################################################################################################################

#   Define a function to extract the barcode sequence from the name of a single file
function getBarcode() {
    local name=`basename "$1"`
    local minbar="$2"
    regex="([ACGT]{$minbar,})"
    [[ $name =~ $regex ]]
    barcode=`echo "${BASH_REMATCH[1]}"`
    echo "$barcode"
}

export -f getBarcode

#   Make sure we have our sample list
if [[ -z "${SAMPLE_INFO}" ]]
then
    echo "Please hard-code a sample list!"
    exit 1
fi

#   Make sure we have our smallest barcode size
if [[ -z "${MIN_BAR}" ]]
then
    echo "Please hard-code a minimum barcode size!"
    exit 1
fi

#   Set an array to hold extracted barcodes
declare -a all_barcodes=()
#   Create a counter to know which index we're storing a barcode into
counter=0

for file in `cat "${SAMPLE_INFO}"` # For each file in our sample list
do
    code=`getBarcode "$file" "${MIN_BAR}"` # Get the barcode sequence
    all_barcodes["$counter"]="$code" # Add it to the barcode array
    let "counter += 1" # Increment the counter
done

declare -a barcode_list=(`tr ' ' '\n' <<< "${all_barcodes[@]}" | sort -u | tr '\n' ' '`) # Create an array of unique barcode sequences

#   Are we missing a header definition?
if [[ -z "${HEADER}" ]]
then
    HEADER="Adapter" # If so, default to "Adapter"
fi

#   Are we missing a name for our output file?
if [[ -z "${OUT}" ]]
then
    #   If so,
    TIME=`date +%m-%d-%y-%H.%M.%S` # What is the date and time?
    OUT="`pwd`/extracted_adapters_${TIME}.txt" # Create a timestamped name for the output file
fi

#   Write each adapter sequence to the output file
for count in `seq 0 "$(( ${#barcode_list[@]} - 1 ))"`
do
    echo -e ">${HEADER}$count\n${PRECEDING_SEQ}${barcode_list[$count]}${FOLLOWING_SEQ}" >> "${OUT}"
done

#   Tell us where the output file is
echo "Adapter file can be found at ${OUT}"