update

misc-scripts/protein_match/get_Xmapping.pl

@@ -217,9 +217,7 @@ if ($organism eq "human") {
 	
 #Its a curated Refseq, flag it as predicted
 	print OUT "$prot_ac\tRefSeq_pred\t$prot_ac\tRefSeq_pred\t$prot_ac\t\tPRED\n";
-       
-	my ($mim) = $_ =~ /\/db_xref=\"MIM:(\d+)/;
-	my ($locus) = $_ =~ /\/db_xref=\"LocusID:(\d*)/;
+
     }
     close (REFSEQ);
     
@@ -232,7 +230,8 @@ if ($organism eq "human") {
     while (<GO>) {
 	chomp;
 	my @array = split (/\t/,$_);
-	print OUT "$array[1]\tSPTR\t$array[4]\tGO\t$array[4]\t\tXREF\n";	
+	$array[9] =~ s/\'/\\\'/g;
+	print OUT "$array[1]\tSPTR\t$array[4]\tGO\t$array[4]\t$array[9]\tXREF\n";	
 	}
 
 }

misc-scripts/protein_match/maps2db.pl

@@ -78,11 +78,11 @@ if ($organism = "human") {
     open (REFSEQPRED,"$refseq_pred") || die "Can't open $refseq_pred\n";
     #Read the file by genbank entries (separated by //) 
     $/ = "\/\/\n";
-    while (<REFSEQ>) {
+    while (<REFSEQPRED>) {
 #This subroutine store for each NP (refseq protein accession number) its corresponding NM (DNA accession number)
 	my ($prot_ac) = $_ =~ /ACCESSION\s+(\S+)/;
 	my ($dna_ac) = $_ =~ /DBSOURCE    REFSEQ: accession\s+(\w+)/;
-
+	#print STDERR "PROT: $prot_ac\t$dna_ac\n";
 	$ref_map_pred{$prot_ac} = $dna_ac;
     }
 #Put back the default (new line) for reading file
@@ -138,6 +138,7 @@ while (<XMAP>) {
 	
 	    ($xac) = $xac =~ /^(XP_\d+)/;
 	    $xac = $ref_map_pred{$xac};
+	    #print STDERR "XAC: $xac\n";
 	}
 
     if ($xid =~ /^XP_\d+/) {
@@ -181,7 +182,7 @@ MAPPING: while (<MAP>) {
     
     my $m = $tid; 
     
-    print STDERR "$queryid,$tid,$tag,$queryperc,$targetperc\n";
+    #print STDERR "$queryid,$tid,$tag,$queryperc,$targetperc\n";
 
     if ($tid =~ /^NP_\d+/) {
 	
@@ -198,7 +199,9 @@ MAPPING: while (<MAP>) {
     if ($tid =~ /^(\w+-\d+)/) {
 	($tid) = $tid =~ /^(\w+)-\d+/;
     }
-    
+ 
+    #print STDERR "TID: $tid\n";
+   
     if ((defined $tid) && (defined $map{$tid})) {
 	
 	

misc-scripts/protein_match/prepare_proteome.pl

@@ -58,6 +58,8 @@ my $pmatch   = $conf{'pmatch'};
 my $organism = $conf{'organism'};
 my $refseq_pred = $conf{'refseq_pred_fa'};
 
+$protfile = "test.fa";
+my $protfile1 = "test_pred.fa";
 
 if (($organism eq "human") || ($organism eq "mouse")) {
     &parse_refseq;
@@ -67,7 +69,7 @@ if ($organism eq "human") {
     &parse_refseq_pred;
 }
 
-&parse_sptr;
+#&parse_sptr;
 &test_protfile;
 
 
@@ -104,7 +106,7 @@ sub parse_sptr {
 sub parse_refseq {
 
   open (IN, "<$refseq") or die "Can't open $refseq\n";
-  open (OUT, ">$protfile") or die "Can't open $protfile\n";
+  open (OUT, ">>$protfile") or die "Can't open $protfile\n";
 
   while(<IN>){
     # eg >gi|4501893|ref|NP_001094.1| actinin, alpha 2 [Homo sapiens]
@@ -130,7 +132,7 @@ sub parse_refseq {
 sub parse_refseq_pred {
     
   open (IN, "<$refseq_pred") or die "Can't open $refseq_pred\n";
-  open (OUT, ">$protfile") or die "Can't open $protfile\n";
+  open (OUT, ">>$protfile1") or die "Can't open $protfile\n";
 
   while(<IN>){
     # eg >gi|4501893|ref|NP_001094.1| actinin, alpha 2 [Homo sapiens]