#!/bin/bash
# THESEUS: Maximum likelihood superpositioning of multiple 
#          macromolecular structures
# 
# -/_|:|_|_\- 
# Douglas L. Theobald
# Department of Biochemistry
# Brandeis University
# Waltham, MA  02454
#
# dtheobald@brandeis.edu
# dtheobald@gmail.com
#
# theseus_align
#
# Usage: theseus_align [theseus_options] -f pdbfile1.pdb pdbfile2.pdb ...
#
# 'theseus_align' allows for a quick-and-dirty way to ML superposition proteins
# with different sequences. It should work very well when the protein sequences
# are relatively similar, although the ML method will still give much better
# results than least-squares when the sequences are moderately divergent.
# Technically, this procedure gives a structure-based superposition of a
# sequence-based alignment. It _does not_ perform a structure-based alignment.
#
# First, the script uses THESEUS to create FASTA formatted sequence files
# corresponding to the exact protein sequences found in the pdb files that you
# supply.
#
# Second, these sequences are aligned using the multiple sequence alignment
# program of your choice -- currently set up for MUSCLE and easily modified for
# CLUSTALW, T_COFFEE, KALIGN, DIALIGN2, or MAFFT. Any multiple sequence
# alignment program can be used, as long as it can generate clustal-formatted
# files. However, I highly recommend Bob Edgar's MUSCLE program for both its
# speed and accuracy. It is easy to install using either precompiled binaries or
# by compiling from scratch:
# http://www.drive5.com/muscle/
#
# Third, THESEUS performs a superposition of the structures using the sequence
# alignment as a guide.
#
# The following six constant strings should be modified to whatever is
# convenient and applicable.

CAT="/bin/cat";
SED="/usr/bin/sed";
TEE="/usr/bin/tee";
theseus="/usr/local/bin/theseus"; # where to find the THESEUS binary executable
fastafile="theseus.fasta";
filemapfile="theseus.filemap";
alignmentfile="theseus.aln";

# for MUSCLE -- http://www.drive5.com/muscle/
alignprog="/usr/local/bin/muscle";
align_cmd="${alignprog} -maxiters 32 -in ${fastafile} -out ${alignmentfile} -clwstrict";

# for PROBCONS -- http://probcons.stanford.edu/
# alignprog="/usr/local/bin/probcons";
# align_cmd="( ${alignprog} -clustalw ${fastafile} | ${SED} 's/PROBCONS/CLUSTALW/' | ${TEE} ${alignmentfile} )";
# echo $align_cmd

# for CLUSTAL-OMEGA -- http://www.clustal.org/omega/
# alignprog="/usr/local/bin/clustalo";
# align_cmd="${alignprog} --force --output-order=input-order -i ${fastafile} -o ${alignmentfile} -v";


# for CLUSTALW -- http://www.clustal.org/clustal2/
#alignprog="/usr/local/bin/clustalw";
#align_cmd="${alignprog} -outorder=input -infile=${fastafile} -outfile=${alignmentfile}";

# for MAFFT -- http://www.biophys.kyoto-u.ac.jp/%7Ekatoh/programs/align/mafft/
# alignprog="/usr/local/bin/mafft";
# align_cmd="${alignprog} --maxiterate 1000 --localpair --clustalout ${fastafile} > ${alignmentfile}";

# for T_COFFEE -- http://igs-server.cnrs-mrs.fr/%7Ecnotred/Projects_home_page/t_coffee_home_page.html
#alignprog="/usr/local/bin/t_coffee";
#align_cmd="${alignprog} ${fastafile} -outfile=${alignmentfile}";

# for KALIGN -- http://msa.cgb.ki.se/
#alignprog="/usr/local/bin/kalign"
#align_cmd="${alignprog} -i ${fastafile} -f aln | sed 's/Kalign/CLUSTALW/' > ${alignmentfile}";

# for DIALIGN2 -- http://bibiserv.techfak.uni-bielefeld.de/dialign/
#alignprog="/usr/local/bin/dialign2"
#align_cmd="${alignprog} -cw ${fastafile}; sed 's/\/\///' ${fastafile%.*}.cw | sed 's/DIALIGN/CLUSTALW/'";


################################################################################
################################################################################
# NOTHING BELOW HERE SHOULD BE CHANGED
################################################################################

if [ ! -f ${alignprog} ] || [ ! -x ${alignprog} ] || [ ! -s ${alignprog} ]
then
    printf "\nERROR: Problem with multiple sequence aligment executable, ${alignprog}\n";
    ${alignprog};
    exit 1;
elif [ ! -f ${theseus} ] || [ ! -x ${theseus} ] || [ ! -s ${theseus} ]
then
    printf "\nERROR: Problem with THESEUS executable, ${alignprog}\n";
    ${theseus};
    exit 1;
fi

usage="Usage: ${0} [theseus_options] -f pdbfile1.pdb pdbfile2.pdb ...";

declare -a opts=( $@ ); # save the command line arguments
(( argc = $# )); # save the number of command line arguments

# shift up until we get past '-f', which signifies that the rest are files
(( optn = 0 ));
while [ "${1}" != "-f" ] && [ ! -z "${1}" ]
do
    shift;
    (( optn++ ));
done

# Ensure that there is something on the command line
if [[ -z "$@" ]] # double brackets don't do word splitting
then
    printf "\n${usage}\n\n";
    exit 1;
else
    shift;
fi

for (( i = optn; i < argc; ++i ))
do
    unset opts[${i}]; # nix the pdb files from the options array
done

pdbs="$@";
echo "PDBs for superpositioning: ${pdbs}"

# Make sure there are pdb files on the command line
if [[ -z "${pdbs}" ]] # double brackets don't do word splitting
then
    printf "\n${usage}\n\n";
    exit 1;
else
    shift;
fi

# Check each pdb file to see if it exists, if it is readable, and if it is non-empty
for pdb in ${pdbs}
do
    if [ ! -f ${pdb} ] || [ ! -r ${pdb} ] || [ ! -s ${pdb} ]
    then
        printf "\nProblem with file: ${pdb}\n"
        printf "\n${usage}\n\n";
        exit 1;
    fi
done

theseus_cmd="${theseus} ${opts[@]} -f -M ${filemapfile} -A ${alignmentfile} ${pdbs}";

# Use THESEUS to make fasta sequence files corresponding to each pdb
${theseus} -f -F ${pdbs};

if [ ! $? ]
then
    printf "\nERROR: THESEUS did not successfully create all FASTA sequence files.\n";
	printf "\n${usage}\n\n";
	exit 1;
fi

for pdb in ${pdbs}
do
    fasta="${pdb}.fst";
    if [ ! -f ${fasta} ] || [ ! -r ${fasta} ] || [ ! -s ${fasta} ]
    then
        printf "\nProblem with FASTA sequence file ${fastafile} for ${pdb}\n"
        printf "\n${usage}\n\n";
        exit 1;
    fi
done

if [ -f ${fastafile} ]
then
    rm ${fastafile};
fi

# Concatenate all fasta files into one large multiple sequence fasta file
for pdb in ${pdbs}
do
    ${CAT} ${pdb}.fst >> ${fastafile};
done

#ls -1 ${pdbs} | awk '{print $1" "$1}' > ${filemapfile};

if [ -f ${filemapfile} ]
then
    rm ${filemapfile};
fi

# Make the mapfile for THESEUS to use (which sequence corresponds to which file)
for pdb in ${pdbs}
do
    echo "${pdb} ${pdb}" >> ${filemapfile};
done

# Align the sequences
printf "\n\n${align_cmd}\n";
eval ${align_cmd};

if [ ! $? ]
then
    printf "\nERROR: Sequence alignment failed.\n";
	printf "\n${usage}\n\n";
	exit 1;
fi


if [ ! -s ${alignmentfile} ]
then
    printf "\nERROR: Sequence alignment failed: no alignment file.\n";
    printf "\n${usage}\n\n";
    exit 1;
fi

# Superimpose with THESEUS based on the sequence alignment generated above
printf "\n\n${theseus_cmd}\n";
${theseus_cmd};

exit 0;