acgt.sh

#!/bin/bash

function die {
  echo -e "\e[31mERROR:\e[0m $2" >&2
  exit $1
}

function usage {
  echo "USAGE: $(basename $0) [-r] FILE"
  echo "       $(basename $0) [-r] - <<<SEQUENCE"
  echo "       $(basename $0) -R STRING"
}

function debug {
  if [ "$DEBUG_MODE" = "1" ]; then
    echo -e "\e[35mDEBUG: $*\e[m" >&2
  fi
}

function acgt-conv {
  tr -d '[:space:]' | while \
    # Read 4 chars from stdin (cut out whitespace first)
    read -r -N 4 BYTE ; do

    # Prepend with trailing T's (0s) from $FILL
    # For convenience we're doing little-endian here
    FILL=tttt
    BYTE="${BYTE}${FILL:${#BYTE}}"

    VAL=0
    for (( i=0 ; i < ${#BYTE} ; i++ )); do
      g=${BYTE:$i:1}
      # A=1 C=2 G=3 T=0
      case $g in
        A|a)  VAL=$(( $VAL + (1 << ($i * 2)) )) ;;
        C|c)  VAL=$(( $VAL + (2 << ($i * 2)) )) ;;
        G|g)  VAL=$(( $VAL + (3 << ($i * 2)) )) ;;
        T|t)  ;;
        *)    die 2 "Format Error: Invalid DNA '$g' in byte '$BYTE'" ;;
      esac
    done

    if (( $VAL > 255 )); then
      die 3 "Math Error: Byte '$BYTE' produced out-of-range value '$VAL'"
    fi

    # Convert decimal number to character byte (in whatever encoding, only ASCII values matter here anyways)
    # It's around here I realize this probably should've been a C program...
    printf "\x$(printf %x $VAL)"
    # Testing: GTGA (little-endian) == AGTG (big-endian) == 115 == 0x73 == 's'
  done
}

# string to DNA
function acgt-reverse {
  while read -rs -n 1 CHAR ; do
    # Convert char to hex, then hex to decimal
    HEX=$(printf "$CHAR" | od -An -t x1 | tr -d ' ')
    VAL=$(( 0x$HEX ))

    if (( $VAL > 255 )); then
      die 2 "Format Error: Character '$CHAR' produced multiple bytes. Only ASCII characters are supported."
    fi

    DNA=
    for (( i=0 ; i < 4 ; i++ )); do
      REM=$(( $VAL % 4 ))
      # A=1 C=2 G=3 T=0
      case $REM in
        1)  DNA="${DNA}A" ;;
        2)  DNA="${DNA}C" ;;
        3)  DNA="${DNA}G" ;;
        0)  DNA="${DNA}T" ;;
      esac
      VAL=$(( $VAL >> 2 ))
    done

    printf "$DNA "
  done
}

if [ -z "$1" ]; then
  usage >&2
  exit 1
elif [[ "$1" = "-h" || "$1" = "--help" ]]; then
  usage
  exit 0
elif [[ "$1" = "-T" || "$1" = "--test" ]]; then
  DEBUG_MODE=1
  shift
elif [[ "$1" = "-R" || "$1" = "--reverse" ]]; then
  REVERSE_MODE=1
  shift
elif [[ "$1" = "-r" || "$1" = "--raw" ]]; then
  OUTPUT=cat
  shift
else
  OUTPUT=strings
fi

if [ "$1" = "-" ]; then
  DNA_FILE=/dev/stdin
else
  DNA_FILE="$1"
fi

if [ "$REVERSE_MODE" = "1" ]; then
  printf "$*" | acgt-reverse
  echo
elif [ "$DEBUG_MODE" != "1" ]; then
  cat $DNA_FILE | acgt-conv | $OUTPUT
fi