# This is part of the parser for MIPS PPI data 

print OUT "bibref\tinteractionDetection\tproteinParticipant\n" ;

@lines = <DATA> ;
chomp(@lines);
$line = join("", @lines) ;
@array = split(/<interaction>/,$line);
&setOut;
for ($i=1 ; $i<@array; $i++){
	while( $array[$i]=~/<bibref>(.*)<\/bibref>/g ){
		$tmp1 = $1 ;
		while($tmp1=~/<xref>\s+<.*db=\"(\w+)\" id=\"(\w+)\".*<\/xref>/g){
			if( $outFile{"experimentDescription"} eq "NA" ){
				$outFile{"experimentDescription"} = $1.":".$2 ;
			}
			else{
				$outFile{"experimentDescription"} = $outFile{"experimentDescription"}.";".$1.":".$2 ;
			}
		}
	}
	while($array[$i]=~/<interactionDetection>(.*)<\/interactionDetection>/g){
		$tmp1 = $1 ;
		while($tmp1=~/<names>\s+<shortLabel>(.*)<\/shortLabel>\s+<\/names>/g){
			if( $outFile{"interactionDetection"} eq "NA" ){
				$outFile{"interactionDetection"} = $1 ;
			}
			else{
				$outFile{"interactionDetection"} = $outFile{"interactionDetection"}.";".$1 ;
			}
		}
	}
	if($array[$i]=~/<participantList>(.*)<\/participantList>/){
		
		@tmp = split(/<proteinParticipant>/,$1) ;
		for($tmp1=1 ; $tmp1<@tmp; $tmp1++){
			$tmp2 = $tmp[$tmp1] ;
			$tmp3 = "" ;			
			if($tmp2=~/<fullName>(.*)<\/fullName>/){
				$tmp3 = $1 ;
				$tmp3 =~s/[\t;]/, /g ;
			}
			if($tmp2=~/<xref>\s+<.*db=\"(\w+)\" id=\"(\w+)\".*<\/xref>/){
				$tmp3 = $1.":".$2."@".$tmp3 ;
			}else{
				$tmp3 = ""."@".$tmp3 ;
			}
			if($tmp2=~/<organism ncbiTaxId="(\d+)" \/>/){
				$tmp3 = $tmp3."@".$1 ;
			}else{
				$tmp3 = $tmp3."@"."" ;
			}
			if( $outFile{"proteinInteractor"} eq "NA" ){
				$outFile{"proteinInteractor"} = $tmp3 ;
			}			
			else{
				$outFile{"proteinInteractor"} = $outFile{"proteinInteractor"}.";".$tmp3 ;
			}
		}
	}
	&writeOut ;
    	&setOut ;
}
close DATA;
close OUT;

sub writeOut {
	print OUT $outFile{"experimentDescription"}, "\t", 
	$outFile{"interactionDetection"}, "\t",  
	$outFile{"proteinInteractor"}, "\n" ;
}
sub setOut{
	$outFile{"experimentDescription"} = "NA";
	$outFile{"interactionDetection"} = "NA";
	$outFile{"proteinInteractor"} = "NA";
}