lociq.script

#!
#########################################################
# Help                                                  #
#########################################################
Help()
{
   # Display Help
   echo "Add description of the script functions here."
   echo
   echo "Syntax: scriptTemplate [-g|h|v|V]"
   echo "options:"
   echo "H     Print this Help."
   echo "a     Path containing analysis files"
}

#########################################################
# PROGRAM						  #
#########################################################

while getopts ":Ha:" opt; do
   case $opt in
      H) # Help info
         Help
         exit;;
      a) # Path to analysis files
	 PATHWAYMR="$OPTARG";;
      \?) #Invalid option
      echo "Option does not exist"
      exit;;
   esac
done

# Navigate to appropriate directory
cd $PATHWAYMR

# Extract sequences from the PLSDB database using ID and range info stored in the 'batcher' file
blastdbcmd -db db.indexed -entry_batch batcher > seqs.1

# Remove range info from headers
sed 's/:.*//' seqs.1 > seqs.2

# Convert each sequence into sets of 2 lines, first being header second being sequence
awk '/^>/ {printf("\n%s\n",$0);next; } { printf("%s",$0);}  END {printf("\n");}' < seqs.2 > seqs.3

# Convert from multifasta sequence to col1 col2 format
awk 'BEGIN{RS=">"}{print $1"\t"$2;}' seqs.3 | tail -n+2 > seqs.4.R

# Cleanup
rm seqs.1
rm seqs.2
rm seqs.3