gene2ptbp1

#!/bin/bash
#PBS -l nodes=1:ppn=4

GENOME="mm9"
BASED=0

#### usage ####
usage() {
    echo
	echo Program: "gene2ptbp1 (retrieve nucleotide sequences corresponding to genomic coordinate(s))"
	echo Author: BRIC, University of Copenhagen, Denmark
	echo Version: 1.0
	echo Contact: pundhir@binf.ku.dk
	echo "Usage: gene2ptbp1 -i <file> -j <file> [OPTIONS]"
	echo "Options:"
    echo " -i <coor>   [input genomic coordinate file in BED format (can be stdin)]"
    echo " -j <file>   [reference genome sequence as fasta file]"
    echo "[OPTIONS]"
    echo " -k <int>    [the coordinate is 0-based or 1-based (default: 0)]"
    echo "             [0: BED and BAM]"
    echo "             [1: GFF, GTF, SAM and VCF]"
    echo "             [for details refer to https://www.biostars.org/p/84686/]"
    echo " -s          [use strand information]"
	echo " -h          [help]"
	echo
	exit 0
}

#### parse options ####
while getopts i:j:k:sh ARG; do
	case "$ARG" in
        i) COORFILE=$OPTARG;;
        j) FASTAFILE=$OPTARG;;
        k) BASED=$OPTARG;;
        s) STRAND=1;;
		h) HELP=1;;
	esac
done

<<"COMMENT"
## usage, if necessary file and directories are given/exist
if [ ! -f "$FASTAFILE" -o -z "$COORFILE" -o "$HELP" ]; then
	usage
fi

###################
#helperfunction
function wait_for_jobs_to_finish {
    for job in `jobs -p`
    do
        echo $job
        wait $job
    done
    echo $1
}
###############

## create temporary BED file if input is from stdin
if [ "$COORFILE" == "stdin" ]; then
    TMP=$(cat /dev/urandom | tr -dc 'a-zA-Z0-9' | fold -w 32 | head -n 1)
    while read LINE; do
        echo -e "${LINE}"
    done > $TMP
    COORFILE=$TMP
fi

## remove temporary file if exists
if [ ! -z "$TMP" ]; then
    rm $TMP
fi
COMMENT

zless analysis/5UTRs.fasta | head -n 10 | perl -ane '
    if($_=~/^>/) {
        chomp($_);
        $ID=$_;
        $ID=~s/\>//g;
    } else {
        print "$ID";
        #while($_=~m!(TCT|TTT|TTC|CTT|CTC|CTG|TGT|CCT|TTG|TCC|GTT|GTC|TGG|GCT|TGC|GTG|GGT|CCC|GCC|GGC[CT]{0,})!g) {
        while($_=~m!(TCT|TTT[CT]{0,})!g) {
            #$matchInfo=sprintf("%s\t%d\t%d", $1, length($`), length($&));
            #$PTBP1{$ID}{'motif'}=$1;
            #$PTBP1{$ID}{'motifStart'}=length($`);
            #$PTBP1{$ID}{'motifLength'}=length($&);
            printf("\t%s,%d,%d", $1, length($`), length($&)); 
        }
        print "\n";
    }'