#!/bin/bash
#
# This is a utility as part of the InterProScan v5 installation
# get nucleotide input and generate ORFs using EMBOSS getorf
#
#usage: translate -i <input_file> -o <output_file>
#

while [[ $# -gt 1 ]]
do
key="$1"

case $key in
    -i|--input)
    INPUTPATH="$2"
    shift # past argument
    ;;
    -o|--output)
    OUTPUTPATH="$2"
    shift # past argument
    ;;
    --default)
    DEFAULT=YES
    ;;
    *)
            # unknown option
    ;;
esac
shift # past argument or value
done
#echo INPUTPATH  = "${INPUTPATH}"
#echo OUTPUTPATH     = "${OUTPUTPATH}"

input_file=`realpath ${INPUTPATH}`
output_file=`realpath ${OUTPUTPATH}`

if [ -f $input_file ]
  then 
  #echo "input is $input_file";
  timenow=`date +"%T"`
  echo "started translate at $timenow";
else
  echo "input file not found: $input_file ";
  exit 
fi

USER_DIR=$PWD

INSTALL_DIR="${BASH_SOURCE[0]}"
while [ -h "$INSTALL_DIR" ]; do
  cd "$(dirname "$INSTALL_DIR")"
  INSTALL_DIR="$(readlink "$(basename "$INSTALL_DIR")")"
done
cd "$(dirname "$INSTALL_DIR")"
INSTALL_DIR="$(pwd)/"

# set environment variables for getorf
export EMBOSS_ACDROOT=${INSTALL_DIR}
export EMBOSS_DATA=${INSTALL_DIR}

${INSTALL_DIR}/getorf -sf pearson -sequence $input_file -outseq $output_file -minsize 75

count=`grep -c '>' $output_file`
echo "$count ORFs in $output_file"
