now also merges adjacent alignment blocks

misc-scripts/assembly/fix_overlaps.pl

@@ -42,6 +42,9 @@ This script removes overlapping mappings that were generated by the code in
 align_nonident_regions.pl. Mappings are trimmed so that no overlaps are present
 in the assembly table, because such overlaps may break the AssemblyMapper when
 projecting between the two assemblies.
+    
+It also merges adjacent assembly segments which can result from alternating
+alignments from clone identity and blastz alignment.
 
 =head1 RELATED FILES
 
@@ -152,8 +155,33 @@ foreach my $chr ($support->sort_chromosomes($support->get_chrlength(undef, undef
   push @rows, $last;
   
   my $i = 0;
+  my $j = 0;
   
   while ($last and (my $r = $sth->fetchrow_hashref)) {
+    # merge adjacent assembly segments (these can result from alternating
+    # alignments from clone identity and blastz alignment)
+    if ($last->{'asm_end'} == ($r->{'asm_start'} - 1) and
+        $last->{'cmp_end'} == ($r->{'cmp_start'} - 1)) {
+
+      $j++;
+
+      # debug warnings
+      $support->log_verbose('last:   '.sprintf($fmt1, map { $last->{$_} }
+        qw(asm_start asm_end cmp_start cmp_end ori)), 1);
+      $support->log_verbose('this: '.sprintf($fmt1, map { $r->{$_} }
+        qw(asm_start asm_end cmp_start cmp_end ori)), 1);
+
+      # remove last row
+      pop(@rows);
+
+      # merge segments and add new row
+      $last->{'asm_end'} = $r->{'asm_end'};
+      $last->{'cmp_end'} = $r->{'cmp_end'};
+      push @rows, $last;
+
+      next;
+    }
+    
     # look for overlaps with last segment
     if ($last->{'asm_end'} >= $r->{'asm_start'}) {
 
@@ -190,6 +218,7 @@ foreach my $chr ($support->sort_chromosomes($support->get_chrlength(undef, undef
   }
 
   $support->log("Fixed $i mappings.\n", 1);
+  $support->log("Merged $j mappings.\n", 1);
 }
 
 $sth->finish;