LBEEP

#!/usr/bin/perl
###################################################################################
# Author: Saravanan Vijayakumar (brsaran@gmail.com)                               #
# Affiliation: CAS in Crystallography and Biophysics, University of Madras, India #
# Date:	18-05-2016								  #
###################################################################################
use strict;
use Cwd qw(abs_path);
use lib abs_path().'/dependency/module/'; #setting module dependnecy path ! Do not change unless you are sure !
use brsaran::DDE; #requires DDE module developed in this study

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

my $path2WEKA = abs_path().'/dependency/'; #change only if weka.jar is in some other location;

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

#######   DO not change anyhting beyond this line, unless if you are sure ###############


############################# Protein Mode Begins ##########################
my $j=0; 
my ($PFhead, $protein,@Position,$FinalOutPep,@PRO,$Pep_DDE,$temp,$i,$Arff_header,$MODEL,@Result,$Pro_Out); #Variable Decclaration for protein mode
my $Window = $ARGV[0];
my $mode = $ARGV[2];
my $protein = $ARGV[1];
my $PFhead = '';
my $MODEL = 'B';

if($mode eq "pro"){# If protein mode is opted
#	$FinalOutPep = 'Position,|Peptide|,Score'."\n"; #Header for CSV
#	print "\nEnter the Window size to be scanned (6-15):";
#	my $Window = <STDIN>;
	$PFhead =~ s/\s//g; #Store Header
	$protein=~s/\s//g; #Store Protein
				
	if(($Window < 6) && ($Window >15)){
#		print "\nWindow size not between 6 and 15 ! May obtain undesired result !\n"; 
	}
	close (INFILE);
	######## Split the protein into mentioned window size and generating inputs #################

	my $Length_Protein = length($protein);
	for($j=0;$j<($Length_Protein - ($Window-1));$j++){
		$PRO[$j] = substr($protein,$j,$Window);
		$Pep_DDE.= DDE($PRO[$j]).",P\n"; #Compute DDE
		$Position[$j]= ($j+1)."-".($j+$Window);
	}
	$temp = './temp/'.generate_random_string(6).'.arff';
	for($i=1;$i<=400;$i++){
		$Arff_header .= "\@attribute $i numeric\n";
	}
	$Arff_header = "\@relation Epitope\n".$Arff_header.'@attribute LABEL {P,N}'."\n"."\@data\n";
	open(TEMP,">$temp");
	print TEMP $Arff_header.$Pep_DDE;
	close(TEMP);
	$Pep_DDE = "";
	$Pro_Out = Epi_Predict_pep($temp,$path2WEKA,$MODEL);#Make prediction	
	$Pro_Out=~s/^(?:.*\n){1,5}//; #Remove line 1-5
	$Pro_Out=~s/(?:.*\n){1,1}\z//; #Remove last line

	`rm $temp`; #remove ARFF file
	@Result = split("\n",$Pro_Out);
	for($j=0;$j<@Result;$j++){
		if($Result[$j]=~m/\+/g){
			$Result[$j]=substr($Result[$j],-6);
			$Result[$j]=~s/\s//g;
			$Result[$j]= 1-$Result[$j];
		}else{
			$Result[$j]=substr($Result[$j],-6);
		}
		$Result[$j]=~s/\s//g;
		$FinalOutPep.= $Position[$j].",".'|'.$PRO[$j].'|,'. $Result[$j]."\n";
	}
	print $FinalOutPep;
	############# Splitting & Prediction Ends #####################
	
	############## HTML Output ENDS #################################
}
################## Protein Mode Ends ########################
##################### PEPTIDE MODE BEGINS #####################################

my (@FHead,@PEP,@PRO,@Result,$Pep_DDE,$j,$line1,$line2,$temp,$Arff_header,$Pro_Out,$Pep_Out,$FinalOutPep); #Variable decclaration
#$FinalOutPep = 'Sno,ID|Peptide|,Score'."\n"; #Header for CSV
$j=0; #Variable for counting the peptides

if($mode eq "pep"){ #If peptide mode is opted
	#$FinalOutPep = 'Sno,ID|Peptide|,Score'."\n";
	open(INFILE,"$protein");
	while(($line1 = <INFILE>) && ($line2 = <INFILE>)){
		$line1=~s/\s//g;		
		$FHead[$j] = $line1; #Store Header
		$line2=~s/\s//g;		
		$PEP[$j]=$line2; #Store Peptide
		if((length($PEP[$j]))>15){print "\nInput peptide $FHead[$j] has length greater than 15. May obtain undesired result ! ";}
		if((length($PEP[$j]))<6){print "\nInput peptide $FHead[$j] has length less than 6 ! May obtain undesired result !";}
		$Pep_DDE.= DDE($line2).",P\n"; #Compute DDE
		$j++;
	}
	close (INFILE);
	$temp = './temp/'.generate_random_string(6).'.arff';
	for($i=1;$i<=400;$i++){
		$Arff_header .= "\@attribute $i numeric\n";
	}
	$Arff_header = "\@relation Epitope\n".$Arff_header.'@attribute LABEL {P,N}'."\n"."\@data\n";
	open(TEMP,">$temp");
	print TEMP $Arff_header.$Pep_DDE;
	close(TEMP);

	$Pep_DDE=""; #Release memory	

	$Pep_Out = Epi_Predict_pep($temp,$path2WEKA,$MODEL);	#Make prediction
	$Pep_Out=~s/^(?:.*\n){1,5}//; #Remove line 1-5 of weka out
	$Pep_Out=~s/(?:.*\n){1,1}\z//; #Remove last line of weka out
#	`rm $temp`; #remove ARFF file

	
	##########Result Organization################
	@Result = split("\n",$Pep_Out);
	for($j=0;$j<@Result;$j++){
		if($Result[$j]=~m/\+/g){
			$Result[$j]=substr($Result[$j],-6);
			$Result[$j]= 1-$Result[$j];
		}else{
			$Result[$j]=substr($Result[$j],-6);
		}
		$Result[$j]=~s/\s//g;
		$FinalOutPep.= $FHead[$j].',|'.$PEP[$j].'|,'. $Result[$j]."\n";
	}
	print $FinalOutPep;
}
######		PEPTIDE MODE ENDS ###########################