cjBitSeqWithinGene

#!/bin/bash

if [ $# -eq 0 ]
  then
        echo "No arguments supplied"
        echo "Display usage info: cjBitSeqWithinGene --help"
        exit 0
fi

if [ $1 == C ]
then
        echo "ERROR: Two sets of files should be provided, separated by C"
        echo "Display usage info: cjBitSeqWithinGene --help"
        echo "exiting."
        exit 0
fi

if [ $1 == --help ]
then
        echo ""
        echo "........................................................................................................................."
        echo ". Usage: cjBitSeqWithinGene <outputDir> <geneToTrFile> <nCores> <.prob files condition A> C <.prob files condition B>   ."
        echo ".                                                                                                              	      ."
        echo ". Collapsed Gibbs algorithm for Differential Transcript Usage analysis                                                  ." 
        echo ".                                                                                                                       ."
        echo ". Arguments:                                                                                                            ."
        echo ".    <outputDir>: Name of the output directory.                                                                         ."
        echo ".                                                                                                                       ."
        echo ".    <geneToTrFile>: File containing gene to transcripts information.                                                   ."
        echo ".                                                                                                                       ."
        echo ".    <nCores>: Number of parallel threads to use (at least 4, but 8 or 12 is strongly suggested).                       ."
        echo ".                                                                                                                       ."
        echo ".    <.prob files condition A>: Space-separated list of '.prob' files for the first sample.                             ."
        echo ".                                                                                                                       ."
        echo ".    <.prob files condition B>: Space-separated list of '.prob' files for the second sample.                            ."
        echo ".                                                                                                                       ."
        echo ". Note: '.prob' files contain alignment probabilities for each read                                                     ."
        echo ".       and they are computed using the 'parseAlignment' command of Bitseq.                                             ."
        echo ".                                                                                                                       ."
        echo ". Detailed documentation and examples: https://github.com/mqbssppe/cjBitSeq/wiki/DTU-inference                          ."
        echo "........................................................................................................................."
        echo ""
        exit 0
fi

if test -d $1; then echo "Directory exists! Please provide different name."; exit 0; fi

mkdir $1
CWD=$(pwd)
echo "output written to $CWD/$1"
a=`expr 1`
b=`expr 0`
c=`expr 2`


echo "First condition files:"
for i in $*
do
        if [ $i == $1 ]
        then
                a=`expr 2`
        fi

        if [ $i == $2 ]
        then
                a=`expr 2`
                cp $i $1/genesTranscripts.txt
                echo "transcript and gene info file: $i"
        fi

        if [ $i == $3 ]
        then
                a=`expr 2`
                echo "Number of cores: $i"
		> $1/nCores.txt && echo "$i" > $1/nCores.txt
        fi



        if [ $a -eq 1 ]
        then
                if [ $i == C ]
                then
                        a=`expr 2`
                        c=`expr 1`
                else
                        b=$((b+1))
                        echo "  : $i, $b"
                        if [ ! -f $i ]; then
                                echo "ERROR: File not found!"
                                echo "exiting."
                                echo "Display usage info: cjBitSeqWithinGene --help"
                                exit 0
                        fi
                        cd $1
                        ln -s ../$i conditionA_$b.prob
                        cd ..
                fi
        fi
        if [ $i == $1 ]
        then
                a=`expr 1`
        fi
        if [ $i == $2 ]
        then
                a=`expr 1`
        fi
        if [ $i == $3 ]
        then
                a=`expr 1`
        fi
done
a=`expr 1`
b=`expr 0`
echo "Second condition files:"
for i in $*
do
	if [ $i == C ]
	then
		a=`expr 2`
		c=`expr 1`
	fi
	if [ $a -eq 2 ]
	then
		if [ $i != C ]
		then
			b=$((b+1))
			echo "	: $i, $b"
			if [ ! -f $i ]; then
				echo "ERROR: File not found!"
				echo "Display usage info: cjBitSeqWithinGene --help"
				echo "exiting."
				exit 0
			fi
			cd $1
			ln -s ../$i conditionB_$b.prob
			cd ..
		fi
	fi
done

if [ $c -eq 2 ]
then
	echo "ERROR: Two sets of files should be provided, separated by C"
	echo "Display usage info: cjBitSeqWithinGene --help"
	echo "exiting."
	exit 0
fi


cd $1
cpThisCOLLAPSEDWITHIN.bash
echo "getting gene information and generating new files (output written to $CWD/$1/partitionMerge2.Rout)"
time R CMD BATCH partitionMerge2.R
echo "done!"
echo "Running Collapsed MCMC using parallel GNU"
cd tmp/
time ./parallelGNUwithin.bash
echo "Done!"
echo "Estimates written to $CWD/$1/withinGeneEstimates.txt"
echo "Cleaning..."
mv getClustersWithin.Rout ../getClustersWithin.Rout
mv nReads.txt ../nReads.txt
cd ..
rm -r tmp
rm getClustersWithin.R partitionMerge2.R *.prob
echo "        ...all done."
exit 0