create submit script for those mappings that produced problems

misc-scripts/xref_mapping/test_mapping_files.pl

@@ -2,6 +2,7 @@ use strict;
 use warnings;
 
 my $dir = shift;
+my $submit = shift;
 
 opendir(DIR, $dir);
 my @allfiles = readdir DIR;
@@ -13,6 +14,11 @@ closedir DIR;
 
 my %hash;
 my %max;
+my %fasta;
+my $method;
+
+$max{'pep'} = 0;
+$max{'dna'} = 0;
 
 foreach my $file(@allfiles){
   if($file eq '.' || $file eq '..'){
@@ -21,18 +27,29 @@ foreach my $file(@allfiles){
     next;
   }else{
     my $type;
-    if($file =~ /dna/){
+    if($file =~ /dna.fasta/){
+      $fasta{'dna'} = $file;
+      next;
+    }
+    elsif($file =~ /protein.fasta/){
+      $fasta{'pep'} = $file;
+      next;
+    }
+    elsif($file =~ /dna/){
       $type = 'dna';
+      if(!defined($method)){
+	if($file =~ /(.*)_dna_\d+.map/){
+	  $method= $1;
+	}
+      }
     }
     elsif($file =~ /pep/){
       $type = 'pep';
     }
-    elsif($file =~ /fasta/){
-      next;
-    }
     else{
       next;
     }
+
     if($file =~ /_(\d+).map/){
       $hash{$type}{'map'}{$1} = $1;
     }
@@ -48,13 +65,29 @@ foreach my $file(@allfiles){
   }
 }
 
+print "Method $method\n";
+
+if(defined($submit)){
+  open(OUT,">SUBMIT")|| die "could not open SUBMIT output file\n";
+}
+
 
 foreach my $type ('pep', 'dna'){
   print "max ".$type."-->".$max{$type}."\n";
   for (my $i=1; $i <= $max{$type}; $i++){
+    
     if(!defined($hash{$type}{'map'}{$i})){
       print $i,".map\n";
+      if(defined($submit)){
+	print OUT "/usr/local/ensembl/bin/exonerate-0.9.0 xref/xref_0_dna.fasta ";
+	print OUT $dir."/homo_sapiens_dna.fasta --querychunkid ".$i." --querychunktotal ";
+	print OUT $max{$type}." --showvulgar false --showalignment FALSE --ryo ";
+	print OUT '"xref:%qi:%ti:%ei:%ql:%tl:%qab:%qae:%tab:%tae:%C:%s\n"';
+	print OUT ' --model affine:local --subopt no --bestn 1 | grep "^xref" > ';
+	print OUT $dir."/".$method."_".$type."_".$i.".map\n";
+      }
     }
+
     if(!defined($hash{$type}{'err'}{$i})){
       print $i,".err\n";
     }
@@ -63,3 +96,5 @@ foreach my $type ('pep', 'dna'){
     }
   }
 }
+
+