createKinaseSubstrate.pl

#!/usr/bin/perl -w
# Author: Abdullah Kahraman
# Date: 15.10.2011

################################################################################
################################################################################
### Extracts the FASTA sequence out of a PDB file.                           ###
################################################################################
################################################################################

use strict;
use warnings;
use Getopt::Long;

my (
    # variable for parameters which are read in from commandline
    $help,
    $seq,
   );

##############################################################################
### read all needed parameters from commandline ##############################

&GetOptions(
    "help!"      => \$help,    # print this help
    "seq=s"      => \$seq,     # sequence of peptide
) or die "\nTry \"$0 -h\" for a complete list of options\n\n";

##############################################################################

# help
if ($help) {printHelp(); exit}

##############################################################################
### SUBROUTINES
############################################################################## 
sub printHelp {
    # prints a help about the using and parameters of this scripts 
    # (execute if user types commandline parameter -h)
    # param:  no paramaters
    # return: no return value

    my (
	$usage,
	$sourceCode,
	@rows,
	$row,
	$option,
	$scriptInfo,
	$example,
       );

    $usage = "$0 -s sequence\n";


    print "\nUsage: cat substrate | " .  $usage . "\n";

    print "Valid options are:\n\n";
    open(MYSELF, "$0") or
      die "Cannot read source code file $0: $!\n";
    $sourceCode .= join "", <MYSELF>;
    close MYSELF;
    $sourceCode =~ s/^.+?\&GetOptions\(\n//s;
    $sourceCode =~ s/\n\).+$//s;
    @rows = split /\n/, $sourceCode;
    foreach $row (@rows){
        $option = $row;
	$option =~ s/\s+\"//g;
	$option =~ s/\"\s.+\#/\t\#/g;
	$option =~ s/=./\t<value> [required]/;
	$option =~ s/:./\t<value> [optional]/;
	$option =~ s/!/\t<non value> [optional]/;

	$row =~ s/^.*//;
	print "\t";
	printf("%-1s%-30s%-30s\n", "-",$option,$row);

    } # end of foreach $row (@rows)
    print "\n";
    print "Options may be abreviated, e.g. -h for --help\n\n";

    $example  = "$0";
}

################################################################################
# MAIN
################################################################################

if(defined $seq){
    my %aa3 = ("GLY" => "G",
	       "ALA" => "A", 
	       "VAL" => "V",
	       "LEU" => "L",
	       "ILE" => "I",
	       "PHE" => "F",
	       "PRO" => "P",
	       "TRP" => "W",
	       "ASN" => "N",
	       "GLN" => "Q",
	       "MET" => "M",
	       "CYS" => "C",
	       "THR" => "T",
	       "TYR" => "Y",
	       "SER" => "S",
	       "ARG" => "R",
	       "LYS" => "K",
	       "HIS" => "H",
	       "ASP" => "D",
	       "GLU" => "E",
	       "ASX" => "B",
	       "GLX" => "Z");
    my %aa1 = ("G" => "GLY",
	       "A" => "ALA", 
	       "V" => "VAL",
	       "L" => "LEU",
	       "I" => "ILE",
	       "F" => "PHE",
	       "P" => "PRO",
	       "W" => "TRP",
	       "N" => "ASN",
	       "Q" => "GLN",
	       "M" => "MET",
	       "C" => "CYS",
	       "T" => "THR",
	       "Y" => "TYR",
	       "S" => "SER",
	       "R" => "ARG",
	       "K" => "LYS",
	       "H" => "HIS",
	       "D" => "ASP",
	       "E" => "GLU",
	       "B" => "ASX",
	       "Z" => "GLX");
	
    my $i = -1;
    my $preResNo = 0;
    my $preAltLoc = " ";
    while(<>){
	if(/^ATOM/ or /^HETATM/){
	    # 1st renumber substrate residues starting from index 1
	    # 2nd replace alternative locations with default 
	    #     location while incrementing residue number
	    # 3rd adjust residue names to supplied sequence while
	    #     removing all non-main chain atoms
	    my $resNo    = substr($_,22,4);
	    my $altLoc   = substr($_,26,1);
	    my $resName  = substr($_, 17, 3);
	    my $atomName = substr($_, 12, 4);
	    next if($atomName !~ / N  / and
		    $atomName !~ / CA / and
		    $atomName !~ / C  / and
		    $atomName !~ / O  /);

	    if($resNo!=$preResNo){
		$i++;
		$preResNo = $resNo;	     
		$preAltLoc = $altLoc;
	    } else {
		if($altLoc ne $preAltLoc){
		    $i++;
		    $preAltLoc = $altLoc;
		}
	    }
	    substr($_,26,1," ");
	    substr($_,22,4,sprintf("%4s",$i+1));
	    substr($_, 17, 3, $aa1{substr($seq, $i, 1)});
	}
	print $_;
    }
} else {
    print STDERR "\nTry \"$0 -h\" for a complete list of options\n\n";   
}