some minor changes

misc-scripts/protein_match/get_embl_mim_mapping.pl

@@ -19,20 +19,30 @@ my %map;
 open (DBMAP,"$dbmap") || die "Can't open file $dbmap\n"; 
 open (OUT,">$out") || die "Can't open file\n";
 
+print STDERR "Reading DBmap\n";
+
 while (<DBMAP>) {
     chomp;
      my ($mapac,$mapdb) = split(/\t/,$_);
      $map{$mapac} = $mapdb;
 }
 
+print STDERR "Reading SP\n";
+
 my $in1  = Bio::SeqIO->new(-file => $sp, '-format' =>'swiss');
 
 while ( my $seq1 = $in1->next_seq() ) {
     my $ac = $seq1->accession;
     my @dblink = $seq1->annotation->each_DBLink;
     
+    
     foreach my $link(@dblink) {
-	if (($link->database eq "EMBL") || ($link->database eq "MIM")) {
+	if (($link->database eq "EMBL") || ($link->database eq "MIM") || ($link->database eq "PDB")) {
+	    
+	    if (!defined $map{$ac}) {
+		die "Can't map $ac\n";
+	    }
+
 	    print OUT "$map{$ac}\t$ac\t".$link->database."\t".$link->primary_id,"\n";
 	}
     }

misc-scripts/protein_match/get_hugo_mapping.pl

@@ -58,6 +58,8 @@ while (<ENS1>) {
     #Get rid of the annoying carriage return!
     $_ =~ s/\r//g;
     my ($hgnc,$sp,$refseq) = split(/\t/,$_);
+
+
     
     if ($sp) {
 	$en1{$sp} = $hgnc;
@@ -108,10 +110,14 @@ while (<DBMAP>) {
 	    
 	    foreach my $sol (@syn) {
 #print the HUGOs aliases
-		print OUT "$mapdb\t$mapac\tALIAS\t$sol\n";
+		print OUT "$mapdb\t$mapac\tHUGO\t$sol\n";
 	    }
 	    
 	}
+	#if (!defined $en2{$en1{$mapac}}) {
+	 #   print STDERR "$mapac\n";
+	#}
+
     }
 }
 

misc-scripts/protein_match/get_mim_locus.pl

@@ -4,7 +4,7 @@ use strict;
 
 use Getopt::Long;
 
-my ($refseq,$dbmap);
+my ($refseq,$dbmap,$out);
 
 my %map;
 
@@ -19,6 +19,8 @@ open (DBMAP,"$dbmap") || die "Can't open file $dbmap\n";
 open (REFSEQ,"$refseq") || die "Can't open file $refseq\n";
 open (OUT,">$out") || die "Can't open file $out";
 
+print STDERR "Reading dbmap\n";
+
 while (<DBMAP>) {
     chomp;
     my ($mapac,$mapdb) = split(/\t/,$_);
@@ -27,9 +29,11 @@ while (<DBMAP>) {
 }
 
 
-
+#Separate by entry (each entry goes into $_)
 $/ = "\/\/\n";
 
+print STDERR "Reading Refseq file\n";
+
 while (<REFSEQ>) {
     my ($prot_ac) = $_ =~ /ACCESSION\s+(\S+)/;
     my ($dna_ac) = $_ =~ /DBSOURCE    REFSEQ: accession\s+(\w+)/;
@@ -39,10 +43,16 @@ while (<REFSEQ>) {
     my ($locus) = $_ =~ /\/db_xref=\"LocusID:(\d*)/;
 
     if ($mim) {
+	if (!defined $map{$dna_ac}) {
+	    die "can't map $dna_ac\n";
+	}
 	print OUT "$map{$dna_ac}\t$dna_ac\tOMIM\t$mim\n";
     }
 
     if ($locus) {
+	if (!defined $map{$dna_ac}) {
+	    die "can't map $dna_ac\n";
+	}
 	print OUT "$map{$dna_ac}\t$dna_ac\tLOCUS\t$locus\n";
     }
 }
@@ -51,3 +61,5 @@ while (<REFSEQ>) {
 
 
 
+
+

misc-scripts/protein_match/process_pmach.pl

@@ -57,13 +57,13 @@ sub postprocesspmatch {
 #Post process the raw data from pmatch
     if ($db eq $sp) {
 	print STDERR "Postprocessing pmatch for SP mapping\n";
-	open (OUT, ">ens_sp.processed");
-	open (PROC, "ens_sp_rawpmatch");
+	open (OUT, ">ens_sp.processed") || die "Can't open File\n";
+	open (PROC, "ens_sp_rawpmatch") || die "Can't open File\n";
     }
     
     else {
 	print STDERR "Postprocessing pmatch for REFSEQ mapping\n"; 
-	open (OUT, ">ens_refseq.processed");
+	open (OUT, ">ens_refseq.processed") || die "Can't open File\n";;
 	open (PROC, "ens_refseq_rawpmatch") || die "Can't open file ens_refseq_rawpmatch\n";
     }