#!/usr/monash/bin/perl

($#ARGV>=0) || die
   "usage: embl2ace [-D] [-o <output_filename>] <filename>
    if -D is presented, the ace file created will have \"-D <Class>: id\" line
    proceeded to each object \"<Class>: id\", as the \"-D\" flag of ACeDB;
    use - for standard input;
    the default output is on the screen;
    options must be given in the order shown above\n";

if ($ARGV[0] =~ /^-D$/)
{       $VAR_D = 1;
        shift
}
else {  $VAR_D = 0 };

# Open all files for output.
open (INPUT, "<$ARGV[0]");
open (OUTPUT1, ">chr1.ace");
open (OUTPUT2, ">chr2.ace");
open (OUTPUT3, ">chr3.ace");
open (OUTPUT4, ">chr4.ace");
open (OUTPUT5, ">chr5.ace");
open (OUTPUT6, ">chr6.ace");
open (OUTPUT7, ">chr7.ace");
open (OUTPUT8, ">chr8.ace");
open (OUTPUT9, ">chr9.ace");
open (OUTPUT10, ">chr10.ace");
open (OUTPUT11, ">chr11.ace");
open (OUTPUT12, ">chr12.ace");
open (OUTPUT13, ">chr13.ace");
open (OUTPUT14, ">chr14.ace");


$clone_count = 0;  # for the construction of grid data
		   # of positive clones to STS probes.

# Get to first Sequence.
while (<INPUT>){
	if (/^BOT/) { last }
}

# Extract Sequence information.
# See sub initialise for variable interpretations.
SCAN:
while (<INPUT>){
	&initialise;
	$STS = <INPUT>;
	$dummy = <INPUT>;
	$primer1 = <INPUT>;
	while (<INPUT>){
		if ($_ =~ /^1,0/) {last}
		else { $primer1 .= $_; }
	}
	$primer2 = <INPUT>;
	while (<INPUT>){
		if ($_ =~ /^1,0/) {
			last;
		}
		else { $primer2 .= $_; }
	}
	$locus = <INPUT>;
	$dummy = <INPUT>;
	$other_names = <INPUT>;
	$PCR_product_size = <INPUT>;
	$dummy = <INPUT>;
	$dummy = <INPUT>;
	$probe_name = <INPUT>;
	$dummy = <INPUT>;
	$probe_sequence = <INPUT>;
	while (<INPUT>){
		if ($_ =~ /^0,/) {
			$YACs = $_;
			last;
		}
		else { $probe_sequence .= $_; }
	}
	$probe_sequence =~ s/[^catgACTG\n]//g;
	$probe_sequence =~ s/^\s*\n//g;
	$probe_sequence =~ s/\n\s*\n/\n/g;
	$dummy = <INPUT>;
	$chromosome = <INPUT>;
        $chromosome =~ s/^0,//;
        $chromosome =~ s/\s*\n//g;
	$chromosome =~ s/^0+//;
	$dummy = <INPUT>;
	$dummy = <INPUT>;
	while (<INPUT>){
		if ($_ =~ /^1,0/) {
			last;
		}
		else { if ($_ !~ /^\s*\n$/) { 
			    $_ =~ s/\n/ \\\n/g; 
			    $comments .= $_; 
		        }
		}
	}
	$comments =~ s/\s*\\\n$/\n/;
	$entry_date = <INPUT>;
	$dummy = <INPUT>;
	$mod_date = <INPUT>;
	$serial_num = <INPUT>;
	$dummy = <INPUT>;
	$dummy = <INPUT>;
	$well_location = <INPUT>;
	$dummy = <INPUT>;
	$YAC_pool = <INPUT>;
	$dummy = <INPUT>;
	$fragment_locale = <INPUT>;
	$dummy = <INPUT>;
	$YAC_library = <INPUT>;
	while (<INPUT>){
		if ($_ =~ /^1,0/) {
			last;
		}
		else { $YAC_library .= $_; }
	}
	$reaction_cond = <INPUT>;
	$dummy = <INPUT>;
	$probe_def = <INPUT>;
	$dummy = <INPUT>;
	$isolates = <INPUT>;
	$probe_size = <INPUT>;
	$dummy = <INPUT>;
	$dummy = <INPUT>;
	$status = <INPUT>;
	$dummy = <INPUT>;
	$public = <INPUT>;
	$dummy = <INPUT>;
	$provisional = <INPUT>;
	$dummy = <INPUT>;
	while (<INPUT>){
		if ($_ =~ /^0,/) {
			$YAC_insert_sizes = $_;
			last;
		}
		else { $approx_locale .= $_; }
	}
	$dummy = <INPUT>;
	$dummy = <INPUT>;
	$Genbank_ID = <INPUT>;
	$dummy = <INPUT>;
	$rows = <INPUT>;
	$columns = <INPUT>;
	$dummy = <INPUT>;
	$dummy = <INPUT>;
	$Lab_YAC_name = <INPUT>;
	$dummy = <INPUT>;
	$dummy = <INPUT>;
	&print_out;
	if ($dummy =~ /^EOD/){
		&printGrid;
		close(OUTPUT1);
		close(OUTPUT2);
		close(OUTPUT3);
		close(OUTPUT4);
		close(OUTPUT5);
		close(OUTPUT6);
		close(OUTPUT7);
		close(OUTPUT8);
		close(OUTPUT9);
		close(OUTPUT10);
		close(OUTPUT11);
		close(OUTPUT12);
		close(OUTPUT13);
		close(OUTPUT14);
		last SCAN;
	}
}
close (INPUT);

# Set variables' values to empty.
sub initialise {
	$STS = "";               # STS name. 
        $primer1 = "";           # Locus Primer1. 
        $primer2 = "";           # Locus Primer2.
        $locus = "";             # Locus name. 
        $other_names = "";       
        $PCR_product_size = "";  # Locus Length.      
	$probe_name = "";         
        $probe_sequence = "";    # DNA sequence.    
	$YACs = "";		 # YAC number.
        $chromosome = "";        # chromosome number.
        $comments = "";          # Sequence and Locus Remark.
        $entry_date = "";        # Sequence and Locus Entry_date. 
        $mod_date = "";          # Sequence and Locus Modification_date.
        $serial_num = "";          
        $well_location = "";   
        $YAC_pool = "";          # Clone In_pool. 
        $fragment_locale = "";          
	$YAC_library = "";       # Clone name = YAC_library + YAC number.
        $reaction_cond = "";     # Locus React_cond.     
        $probe_def = "";         
        $isolates = "";          # Clone Isolate. 
        $probe_size = "";        # Sequence DNA Int. 
        $status = "";            # Locus Status.
        $public = "";            
        $provisional = "";       # Locus Provisional. 
        $approx_locale = "";     
        $YAC_insert_sizes = "";  # Clone Gel_length. 
        $Genbank_ID = "";     
	$rows = "";     
        $columns = "";     
        $Lab_YAC_name = "";     
}

# Print out instance of Sequence and
# associated instances of Locus, Clones, STS, Oligos, and Grid_data.
sub print_out {
	local($i);

	if ($STS =~ /^\s*""\s*$/){
		return;
	}

	# Print to file that sequence is relevant to.
	# that is, the chromosome number.
	$file_desig = $chromosome;
	if ($file_desig eq 1) { select(OUTPUT1); }
	elsif ($file_desig eq 2) { select(OUTPUT2); }
	elsif ($file_desig eq 3) { select(OUTPUT3); }
	elsif ($file_desig eq 4) { select(OUTPUT4); }
	elsif ($file_desig eq 5) { select(OUTPUT5); }
	elsif ($file_desig eq 6) { select(OUTPUT6); }
	elsif ($file_desig eq 7) { select(OUTPUT7); }
	elsif ($file_desig eq 8) { select(OUTPUT8); }
	elsif ($file_desig eq 9) { select(OUTPUT9); }
	elsif ($file_desig eq 10) { select(OUTPUT10); }
	elsif ($file_desig eq 11) { select(OUTPUT11); }
	elsif ($file_desig eq 12) { select(OUTPUT12); }
	elsif ($file_desig eq 13) { select(OUTPUT13); }
	elsif ($file_desig eq 14) { select(OUTPUT14); }
	else { print "CHROM: $file_desig\n"; }



        # Print out Sequence.	
	$STS =~ s/\s*//g;
	$STS =~ s/\n//;
	print "\n-D Sequence : $STS\n\n" if ($VAR_D);
	print "Sequence : $STS\n";
	$probe_size =~ s/^0,//;
	$probe_size =~ s/[^0-9]//g;
	print "DNA $STS $probe_size\n";
	$entry_date =~ s/\"//g;
	$entry_date =~ s/\n//g;
	$ACE_entry_date = &AceDate($entry_date);
	print "Entry_date \"$ACE_entry_date\"\n"; 
	$mod_date =~ s/\"//g;
	$mod_date =~ s/\n//g;
	$ACE_mod_date = &AceDate($mod_date);
	print "Modification_date \"$ACE_mod_date\"\n"; 
	$YACs =~ s/^0,//;
	$YACs =~ s/\n//;
	$YACs =~ s/[a-zA-Z]//g;
	$YACs =~ s/\s+/ /g;
	@YAC_array = split(" ", $YACs);
	$YAC_library =~ s/\n//g;
	$YAC_library =~ s/^"//g;
	$YAC_library =~ s/"\s*$//g;
	@YAC_lib_array = split('" "', $YAC_library);
	print "Locus $STS\n";
	for ($i = 0; $i <= $#YAC_array; $i++){
		$YAC_array[$i] =~ s/\s/_/g;
		$YAC_lib_array[$i] =~ s/\s/_/g;
		print "Clone \"$YAC_lib_array[$i]_$YAC_array[$i]\"\n";
	}	
	$Oligo1 = $STS;
	$Oligo1 =~ s/"\s*$/:1"/;
	$Oligo2 = $STS;
	$Oligo2 =~ s/"\s*$/:2"/;
	print "Oligo $Oligo1\n";
	print "Oligo $Oligo2\n";
	print "Remark $comments";  
	print "Organism \"Plasmodium falciparum\"\n";
	print "\n\n\n\n";


	# Print out Locus.
	print "-D Locus : $STS\n\n" if ($VAR_D);
	print "Locus : $STS\n";
	$entry_date =~ s/\"//g;
	$entry_date =~ s/\n//g;
	$ACE_entry_date = &AceDate($entry_date);
	print "Entry_date \"$ACE_entry_date\"\n"; 
	$mod_date =~ s/\"//g;
	$mod_date =~ s/\n//g;
	$ACE_mod_date = &AceDate($mod_date);
	print "Modification_date \"$ACE_mod_date\"\n"; 
	$ACE_mod_date = &AceDate($mod_date);
	print "Modification_date \"$ACE_mod_date\"\n"; 
	$primer1 =~ s/\n//g;
	print "Primer1 $primer1\n"; 
	$primer2 =~ s/\n//g;
	print "Primer2 $primer2\n"; 
	print "React_cond $reaction_cond"; 
	$PCR_product_size =~ s/^0,//;
	$PCR_product_size =~ s/[^0-9]//g;
	print "Length $PCR_product_size\n";
	for ($i = 0; $i <= $#YAC_array; $i++){
		print "Positive_clone \"$YAC_lib_array[$i]_$YAC_array[$i]\"\n";
	}	
	print "Sequence $STS\n";
	print "Remark $comments";  
	$map = &stringtoRoman($chromosome);
	if ($STS =~ /-0*([1-9][0-9]*)"/){
#		May multiply this by appropriate scalar
		$map_position = $1;
	}
	print "Map \"$map\" Position $map_position\n";
	print "Status $status";
	print "Provisional $provisional";			
	print "Organism \"Plasmodium falciparum\"\n";
	print "\n\n\n\n";	

	
	# Print out STS.
	print "\n-D DNA : $STS\n\n" if ($VAR_D);
	print "DNA : $STS\n";
	$probe_sequence =~ s/"//g;
	print $probe_sequence;
	print "\n\n\n\n";

	# Print out Clone.
	$YAC_pool =~ s/"//g;		
	$YAC_pool =~ s/\n//g;		
	@In_pool = split(" ", $YAC_pool);
	$isolates =~ s/\n//g;	
	$isolates =~ s/^"//g;	
	$isolates =~ s/"\s*$//g;
	@isolate_array = split('" "', $isolates);
	$YAC_insert_sizes =~ s/^0,//;
	$YAC_insert_sizes =~ s/\s*\n//;
	$YAC_insert_sizes =~ s/[a-zA-Z]//g;	
	$YAC_insert_sizes =~ s/\s+/ /g;	
	@YAC_length_array = split(" ", $YAC_insert_sizes);
	for ($i = 0; $i <= $#YAC_array; $i++){
	 print "\n-D Clone : \"$YAC_lib_array[$i]_$YAC_array[$i]\"\n" if ($VAR_D);
	 print "Clone : \"$YAC_lib_array[$i]_$YAC_array[$i]\"\n";
	 print "YAC\n";
	 print "Sequence $STS\n";
	 print "In_pool $In_pool[$i]\n" if ($i <= $#In_pool);   
	 print "Isolate \"$isolate_array[$i]\"\n" if ($i <= $#isolate_array);
	 print "Gel_length $YAC_length_array[$i]\n" if ($i <= $#YAC_length_array);
	 print "Positive_locus $STS\n";  
	 print "Organism \"Plasmodium falciparum\"\n";
	 print "\n\n\n";	
	}
	
	
	# Print out Oligo1. re. Primer1.
	print "-D Oligo : $Oligo1\n\n" if ($VAR_D);
	print "Oligo : $Oligo1\n";
	print "Sequence $primer1\n";
	print "In_sequence  $STS\n";
	print "STS $STS\n";
	print "\n\n\n";


	# Print out Oligo21. re. Primer2.
	print "-D Oligo : $Oligo2\n\n" if ($VAR_D);
	print "Oligo : $Oligo2\n";
	print "Sequence $primer2\n";
        print "In_sequence  $STS\n";
	print "STS $STS\n";
        print "\n\n\n";


	# Print out STS.
        print "-D STS : $STS\n\n" if ($VAR_D);   
        print "STS : $STS\n";   
        print "Oligo1 $Oligo1\n";
        print "Oligo2 $Oligo2\n";
        $Grid_data = $STS;
        $Grid_data =~ s/"//g;
	print "Grid_data \"Grid_" , $Grid_data , "_data\"\n";
        print "\n\n\n";

	print "-D Grid_data : \"Grid_" , $Grid_data , "_data\"\n\n" if ($VAR_D);
	print "Grid_data : \"Grid_" , $Grid_data , "_data\"\n";
	print "Date \"$ACE_entry_date\"\n"; 
	for ($i = 0; $i <= $#YAC_array; $i++){
	    print "Hybridizes_to \"$YAC_lib_array[$i]_$YAC_array[$i]\"\n";
	}	
	print "Grid \"Grid_" , $Grid_data , "\"\n";
	print "Default_negative\n";
	print "\n\n\n";

	
	#  Store positive clones for each grid that represents an STS probe.
	# This information is used by sub printGrid.

	# Store the name of the grid. 
	$Grid_data[$clone_count] = $Grid_data;
	# Each grid is to be printed to the relevant file.
	$file_desig[$clone_count] = $file_desig;
	# Store the list of positive clones for this STS.
	for ($i = 0; $i <= $#YAC_array; $i++){
	      $clone[$clone_count][$i] = "\"$YAC_lib_array[$i]_$YAC_array[$i]\"";
	}	
	$clone[$clone_count][$i++] = ""; # End of Pos. clones flag.
	$clone_count++;

}

# Print each Grid for each STS to the relevant file.
sub printGrid {

     local($i, $j, $k);
	
     for ($i = 0; $i < $clone_count; $i++){
	# Select file for this grid.
	if ($file_desig[$j] eq 1) { select(OUTPUT1); }
	elsif ($file_desig[$i] eq 2) { select(OUTPUT2);  }
	elsif ($file_desig[$i] eq 3) { select(OUTPUT3);  }
	elsif ($file_desig[$i] eq 4) { select(OUTPUT4);  }
	elsif ($file_desig[$i] eq 5) { select(OUTPUT5);  }
	elsif ($file_desig[$i] eq 6) { select(OUTPUT6); }
	elsif ($file_desig[$i] eq 7) { select(OUTPUT7); }
	elsif ($file_desig[$i] eq 8) { select(OUTPUT8); }
	elsif ($file_desig[$i] eq 9) { select(OUTPUT9);  }
	elsif ($file_desig[$i] eq 10) { select(OUTPUT10); }
	elsif ($file_desig[$i] eq 11) { select(OUTPUT11);  }
	elsif ($file_desig[$i] eq 12) { select(OUTPUT12);  }
	elsif ($file_desig[$i] eq 13) { select(OUTPUT13);  }
	elsif ($file_desig[$i] eq 14) { select(OUTPUT14);  }
	print "-D Grid : \"Grid_" , $Grid_data[$i] , "\"\n\n" if ($VAR_D);
	# Print out instance of Grid.
	print "Grid : \"Grid_" , $Grid_data[$i] , "\"\n";
	print "Title \"Grid_" , $Grid_data[$i] , "\"\n";
	print "Space_at 2 3\n";
	print "View \"Grid_type_1\"\n";
	# Initialise row number.
	$gridrow = 0;
	for ($j = 0; $j < $clone_count; $j++){
	   # Print out rows of all clones, for all STSs, 
	   # pertaining to the present file.
	   if ($file_desig[$i] eq $file_desig[$j]){
		$gridrow++;
		print "Row $gridrow Clone";
		$k = 0;
		while ($clone[$j][$k] ne ""){
			print " $clone[$j][$k]";
			$k++;
		}
		print "\n";
	    }
	}
	print "\n\n\n";
     }
}

		
	
# Changes input like 29/2/97 to 97-2-29 .
sub AceDate {
		@VAR_date = ();
                $_ =~s/\s//g;
                $_ =~s/\n//g;
                @temp_date = split ("/", $_[0]);
                $VAR_date[0] = $temp_date[2];
                $VAR_date[2] = $temp_date[0];
		$VAR_date[1] = $temp_date[1];
                $VAR_date = join("-", @VAR_date);
		return $VAR_date;
};

# Auxillary string to integer function.
sub strtoint{
        local($num, $i, @digits);
        @digits = split("" , $_[0]);
        $num = 0;
        for ($i = 0; $i <= $#digits; $i++){
                $num += (10**($#digits - $i)) * (ord($digits[$i]) - 48);
        }
        return $num;
}

# Auxillary English numeral to Roman numeral for integers < 100.
sub stringtoRoman {
     local($Roman, $Roman2, $i, @digits);
     @digits = ();
     $Roman2 = "";
     $Roman= "";
     @digits = split("" , $_[0]);
     @digits = reverse(@digits);
     if ($#digits >= 0){
	$dig = ord($digits[0]) - 48;
	if ($dig == 9){
		$Roman = "IX";
	} elsif ($dig == 4){
		$Roman = "IV";
	} else {
	  if ($dig > 5){
		$Roman = "V";
	  }
	  $dig = $dig % 5;		
	  for ($i = 0; $i < $dig; $i++){
		$Roman .= "I"; 
	  }
	}
     }
     if ($#digits >= 1){
	$dig = ord($digits[1]) - 48;
	if ($dig == 9){
                $Roman2 = "XC";
        } elsif ($dig == 4){
                $Roman2 = "XL";
        } else {
          if ($dig > 5){
                $Roman2 = "L";           
          }
          $dig = $dig % 5;
          for ($i = 0; $i < $dig; $i++){
                $Roman2 .= "X";
          }
        }
      }
      $Roman2 .= $Roman;
      return $Roman2;
}


exit;