# This is part of the parser for IPI data 

# From KEGG Gene ID to KEGG Pathway
while(<In_PATH>){
	chomp;
	@array = split(/\t/,$_);
	$array[1] =~s/path:// ;
	if( !exists($KEGG_path{$array[0]}) ){
		$KEGG_path{$array[0]} = $array[1] ;
	}
	else{
		$KEGG_path{$array[0]} = $KEGG_path{$array[0]}.";".$array[1]  ;
	}
}
close(In_PATH);

# From NCBI Gene ID to KEGG Gene ID
while(<In_KEGG>){
	chomp;
	@array = split(/\t/,$_) ;
	$array[1] =~s/ncbi-geneid:// ;
	$KEGG{$array[1]} = $array[0] ;
}
close(In_KEGG);


# From NCBI Gene ID to GO
<In_GO>;
while(<In_GO>){
	chomp;
	@array = split(/\t/,$_);
	if (  $organism == $array[0] ){
		if($array[7] eq "Function"){
			if( !exists($go{$array[1]}) ){
				$go{$array[1]} = $array[2]."@".$array[3]."@"."MF" ;
			}
			else{
				$go{$array[1]} = $go{$array[1]}.";".$array[2]."@".$array[3]."@"."MF" ;
			}
		}
		elsif($array[7] eq "Component"){
			if( !exists($go{$array[1]}) ){
				$go{$array[1]} = $array[2]."@".$array[3]."@"."CC" ;
			}
			else{
				$go{$array[1]} = $go{$array[1]}.";".$array[2]."@".$array[3]."@"."CC" ;
			}
		}
		elsif($array[7] eq "Process"){
			if( !exists($go{$array[1]}) ){
				$go{$array[1]} = $array[2]."@".$array[3]."@"."BP" ;
			}
			else{
				$go{$array[1]} = $go{$array[1]}.";".$array[2]."@".$array[3]."@"."BP" ;
			}
		}
	}
}
close(In_GO);

print OUT  "IPIID", "\t", "IPIAC", "\t", "IPIACs", "\t", "LEN", "\t", "MW", "\t", "DE", "\t", "CHR", "\t", 
"START", "\t", "END", "\t", "ORIENT", "\t", 
"SYMBOL", "\t", "SPAC", "\t", "SPID", "\t", "REFSEQ", "\t", "GI", "\t", "GENEID", "\t", "UNIGENE", "\t",
"GO", "\t", "KEGG", "\t", "PATH", "\t", "INTERPRO", "\t", "PFAM", "\t", "PROSITE", "\t", 
"SQ", "\n";

#%outFile;
&setOut;

while($line=<DATA>){
	chomp($line) ;
	if($line=~/^ID\s+([\w\.]+)\s+.*\s(\d+)\s+AA\./){
		$outFile{"ID"} = $1;
		$outFile{"LEN"} = $2;
	}
	if($line=~/^AC\s+(.*);$/){				
		if( $outFile{"ACs"} eq "NA" ){
			$outFile{"ACs"}=$1;
		}
		else{
			$outFile{"ACs"}=$outFile{"ACs"}.";".$1;
		}
	}
	elsif($line=~/^DE\s+(.*)/){
		if( $outFile{"DE"} eq "NA" ){
			$outFile{"DE"}=$1;	
		}else{
			$outFile{"DE"}=$outFile{"DE"}." ".$1;		
		}
	}
	elsif($line=~/^CC\s+\-!\- GENE_LOCATION:.*Chr\.\s+(\d+):(\d+)\-(\d+):([\-\d]+)/){
		$outFile{"CHR"} = $1 ;
		$outFile{"START"} = $2 ; 
		$outFile{"END"} = $3 ;
		if ($4==1){
			$outFile{"ORIENT"} = "+" ;
		}
		elsif ($4==-1){
			$outFile{"ORIENT"} = "-" ;
		} 
	}
	elsif($line=~/^DR\s+REFSEQ_REVIEWED;\s+([\w\.]+);\s+GI:(\d+);/){
		$outFile{"RefSeq"} = $1 ;
		$outFile{"GI"} = $2 ;
	}
	elsif($line=~/^DR\s+UniProtKB\/Swiss-Prot;\s+([\w\-]+);\s+(\w+);/){
		$outFile{"spAC"} =$1 ;	
		$outFile{"spID"} =$2 ;
		$outFile{"spAC"} =~s/\-\d+// ;				
	}
	elsif($line=~/^DR\s+Entrez Gene;\s+(\d+);\s+(\w+);/){				
		$outFile{"GS"} = $2 ;
		$outFile{"GeneID"} = $1 ;
		if( exists($KEGG{$1}) ){
			$outFile{"KEGG"} = $KEGG{$1} ;
		}
		if( exists($KEGG_path{$outFile{"KEGG"}}) ){
			$outFile{"PATH"} = $KEGG_path{$outFile{"KEGG"}} ;
		}
		if( exists($go{$1}) ){
			$outFile{"GO"} = $go{$1} ; 
		}
	}
	elsif($line=~/^DR\s+UniGene;\s+([\w\.]+);/){
		$outFile{"UniGene"} = $1 ;
	}
	elsif($line=~/^DR\s+InterPro;\s+(\w+);/){
		if($outFile{"INTERPRO"} eq "NA"){
			$outFile{"INTERPRO"} = $1;
		}
		else{
			$outFile{"INTERPRO"} = $outFile{"INTERPRO"}.";".$1 ;
		}
	}	
	elsif($line=~/^DR\s+Pfam;\s+(\w+);/){
		if($outFile{"PFAM"} eq "NA"){
			$outFile{"PFAM"} = $1;
		}
		else{
			$outFile{"PFAM"} = $outFile{"PFAM"}.";".$1 ;
		}
	}
	elsif($line=~/^DR\s+PROSITE;\s+(\w+);/){
		if($outFile{"PROSITE"} eq "NA"){
			$outFile{"PROSITE"} = $1;
		}
		else{
			$outFile{"PROSITE"} = $outFile{"PROSITE"}.";".$1 ;
		}
	}
	elsif($line=~/^SQ\s+SEQUENCE.*\s(\d+)\s+MW;/){
		$outFile{"MW"} = $1;
		$tag=1;
	}
	elsif($tag==1){
		if( $outFile{"SQ"} eq "NA" ){
			$outFile{"SQ"}=$line;
		}else{
			$outFile{"SQ"}=$outFile{"SQ"}.$line;
		}
	}
	
	if($line=~/^\/\//){
		$outFile{"ACs"} =~s/\s+//g;	
		@tmp_ACs = split(/;/,$outFile{"ACs"})	;
		$outFile{"AC"} = $tmp_ACs[0] ;	
		$outFile{"SQ"}=~s/[\s\n\/]//g;
		&writeOut ;
    		&setOut ;
	}
}
close DATA;
close OUT;

sub writeOut {
	print OUT $outFile{"ID"}, "\t", 
 	$outFile{"AC"}, "\t",   
 	$outFile{"ACs"}, "\t",
	$outFile{"LEN"}, "\t",  
	$outFile{"MW"}, "\t", 
  $outFile{"DE"}, "\t",
  $outFile{"CHR"}, "\t", 
	$outFile{"START"}, "\t", 
	$outFile{"END"}, "\t", 
	$outFile{"ORIENT"}, "\t",  
  $outFile{"GS"}, "\t",
  $outFile{"spAC"}, "\t",
  $outFile{"spID"}, "\t",
  $outFile{"RefSeq"}, "\t",
  $outFile{"GI"}, "\t",
  $outFile{"GeneID"}, "\t",
  $outFile{"UniGene"}, "\t",
	$outFile{"GO"}, "\t",
	$outFile{"KEGG"}, "\t",
	$outFile{"PATH"}, "\t",
	$outFile{"INTERPRO"}, "\t",
	$outFile{"PFAM"}, "\t",
	$outFile{"PROSITE"}, "\t",	
	$outFile{"SQ"}, "\n" ;
}

sub setOut{	
	$tag=0;	
	$outFile{"ID"} = "NA";
	$outFile{"AC"} = "NA";
	$outFile{"ACs"} = "NA";
	$outFile{"LEN"} = "NA";
	$outFile{"MW"} = "NA";
	$outFile{"DE"} = "NA";
	$outFile{"CHR"} = "NA";
	$outFile{"START"} = "NA";
	$outFile{"END"} = "NA";
	$outFile{"ORIENT"} = "NA"; 
	$outFile{"GS"} = "NA";
	$outFile{"spAC"} = "NA";
	$outFile{"spID"} = "NA";
  $outFile{"RefSeq"} = "NA";
  $outFile{"GI"} = "NA";
  $outFile{"GeneID"} = "NA";
  $outFile{"UniGene"} = "NA";
	$outFile{"GO"} = "NA";
	$outFile{"KEGG"} = "NA";
	$outFile{"PATH"} = "NA";
	$outFile{"INTERPRO"} = "NA";
	$outFile{"PFAM"} = "NA";
	$outFile{"PROSITE"} = "NA";	
	$outFile{"SQ"} = "NA";	
}