BlastPDB

#!/bin/sh
#**************************************************************
#  EPPIC Pre-Computation Script
#  Author : Kumaran Baskaran
#  Date   : 05/07/2013
#***************************************************************


help=" Usage: $0 \n
        \t\t    [ -i <dir>  : Input Directory where the fasta files are present \n
        \t\t      -o <dir>  : Output Directory where blast files are to be written \n
        \t\t      -l <dir>  : Logs Directory\n
        \t\t      -f <file> : Queries file containing list of fasta files to be blasted \n
	\t\t	  -d <dir>  : Uniprot Database Path(Preferably in the local scratch) \n
	\t\t	  -p <path> : Blast Program Path  ] \n
Example: ./blast-the-pdb -i /home/in -o /home/out -l /home/logs/ -f /home/queries.list -d /scratch/biyanin/uniprot_2013_01 -p /bin/blastp "

# Set Defaults
BLASTPBIN=/gpfs/home/duarte_j/software/packages/ncbi-blast-2.2.27+/bin/blastp

while getopts :i:o:l:f:d:p:h option
do
        case "${option}"
        in
                i) blastqueriesdir=${OPTARG};;
                o) outdir=${OPTARG};;
                l) logsdir=${OPTARG};;
                f) queriesfile=${OPTARG};;
		d) BLASTDBDIR=${OPTARG};;
		p) BLASTPBIN=${OPTARG};;
                h) echo -e $help;;
               \?) print >&2 $help
                   exit 1;;
                :) echo "`date +%d/%m/%y-%H:%M:%S` ERROR: Option -$OPTARG requires an argument." >&2
                   exit 1;;
        esac
done

if [ -z $blastqueriesdir ] || [ -z $outdir ] || [ -z $logsdir ] || [ -z $queriesfile ] || [ -z $BLASTDBDIR ]
then
        echo -e "\n `date +%d/%m/%y-%H:%M:%S` ERROR: ---- SOME OPTIONS NOT SPECIFIED CORRECTLY ---- "
        echo -e $help
        exit 1
fi


if [ ! -d $outdir ]
then
	mkdir -p $outdir
fi
if [ ! -d $logsdir ]
then
	mkdir -p $logsdir
fi



blastdb=$BLASTDBDIR/uniref100.fasta

numthreads=1
srt="23:40:00"
hrt="24:00:00"

maxjobs=`cat $queriesfile | wc -l`

cur_dir=`pwd`
date=`date +%Y%m%d`
scriptname=$cur_dir/blast_submit-$date.sh

cat << _EOF_ > $scriptname
#!/bin/sh

#$ -N pdb-blast
#$ -q all.q
#$ -e $logsdir
#$ -o $logsdir
#$ -t 1-$maxjobs
#$ -l ram=8G
#$ -l s_rt=$srt,h_rt=$hrt

query=\`sed "\${SGE_TASK_ID}q;d" $queriesfile\`

chars=\`grep -v "^>" $blastqueriesdir/\$query.fa | wc -c\`
lines=\`grep -v "^>" $blastqueriesdir/\$query.fa | wc -l\`
count=\$(( chars-lines ))

matrix=BLOSUM62

if [ \$count -lt 35 ]; then
	matrix=PAM30
else
	if [ \$count -lt 50 ]; then
	matrix=PAM70
	else
		if [ \$count -lt 85 ]; then
		matrix=BLOSUM80
		fi
	fi
fi

time $BLASTPBIN -matrix \$matrix -db $blastdb -query $blastqueriesdir/\$query.fa -num_threads $numthreads -outfmt 5 -seg no | gzip > $outdir/\$query.blast.xml.gz
_EOF_

chmod +x $scriptname
echo "`date +%d/%m/%y-%H:%M:%S` INFO: $scriptname is created"
#echo "Submitting Job to server.."
#qsub $scriptname