filterL6.sh

#!/bin/bash
usage ()
{
    echo ""
    echo "Usage : `basename $0` <tableL6.txt> <L6_clean.tsv>"
    echo ""
    exit
}

if [ "$1" = "" ]; then
    usage
fi


if [ "$2" != "" ]; then

  cat $1 | head -2 | tail -1 | sed 's/ /_/; s/#//' | perl -pe 's/$/\tSum/' > fx_ids_rel

  cat $1 | sed 's/\[\|\]//g; s/ /_/g; s/k__//' | grep "g__[a-zA-Z]" | \
                               perl -pe 's/\;[pcofg]\_\_/\|/g' > fx_genus_asssigned_rel

  cat $1 | grep -v \# | sed 's/\[\|\]//g; s/ /_/g' | grep -v "g__[a-zA-Z]" | \
      sed 's/;c__;o__;f__;g__//; s/;o__;f__;g__//; s/;f__;g__//; s/;g__\t/\t/; s/k__//' | \
      perl -pe 's/\;[pcofg]\_\_/\|/g; s/\;Other{1,}/\+/g' >  fx_genus_unasssigned_rel

  cat fx_genus_asssigned_rel fx_genus_unasssigned_rel | sort | \
      awk 'BEGIN {FS=OFS="\t"} { sum=0; n=0 ;for(i=3;i<=NF;i++) {sum+=$i; ++n} print $0, sum }' |\
      awk -F'\t' '$NF!="0"'  | grep -v Unassigned > fx_sum_rel

  cat fx_ids_rel fx_sum_rel  > fx_tmp_rel

  awk -v OFS='\t' -F'\t' 'NF{NF--};1' < fx_tmp_rel > $2 && rm fx*rel

else
	echo ""
	echo "Parameter Not Valid!"
  exit 0
fi