Changeset 249

Timestamp:

08/31/09 17:07:51 (3 months ago)

Author:

cholt

Message:

index error fixes

Files:

• bin/maker (modified) (2 diffs)
• lib/FastaDB.pm (added)
• lib/GI.pm (modified) (19 diffs)
• lib/Widget/RepeatMasker.pm (modified) (1 diff)
• lib/Widget/blastn.pm (modified) (1 diff)
• lib/Widget/blastx.pm (modified) (1 diff)
• lib/Widget/tblastx.pm (modified) (1 diff)

bin/maker

@@ -58 +58 @@
-use Getopt::Long qw(:config no_ignore_case);
-use File::Temp qw(tempfile tempdir);
-
+#
-use GI;
-use Dumper::GFF::GFFV3;
@@ -253 +253 @@
-
-    next if($tier->terminated);
- while(! $tier->terminated && ! $tier->failed)
+_all
-    $DS_CTL->add_entry($tier->DS);
-    push(@failed, $tier->fasta) if ($tier->failed);

lib/GI.pm

@@ -34 +34 @@
-use PhatHit_utils;
-use Shadower;
-
+#
-use polisher::exonerate::protein;
-use polisher::exonerate::est;
@@ -42 +42 @@
-use maker::sens_spec;
-use File::NFSLock;
+use FastaDB;
+use Digest::MD5;
-use threads;
-
@@ -213 +215 @@
-      #search db index
-      my $fastaObj = $db_index->get_Seq_by_id($hit->name);
+      $fastaObj = $db_index->get_Seq_by_alias($hit->description) if(not $fastaObj);
+      
+      #still no sequence? try rebuilding the index and try again
-      if (not $fastaObj) {
-
-
-
-
-
-
-
-
-
+
+
+
+
+
+
+
+
-      }
-      
@@ -228 +232 @@
-      my $t_seq      = $fastaObj->seq();
-      my $t_def      = $db_index->header($hit->name);
-
+
+
-      #write fasta file
-      my $fasta = Fasta::toFasta('>'.$t_def, \$t_seq);
@@ -520 +525 @@
-   my $mpi_size = shift @_ || 1;
-
+   #always set to at least 10 for faster fasta indexing
+   $mpi_size = 10 if($mpi_size < 10);
+
-   my $file = $CTL_OPT->{$key};
-   my $alt = $CTL_OPT->{alt_peptide} if($key =~ /protein/);
@@ -530 +538 @@
-   my $bins = $mpi_size;
-   $bins = $db_size if ($db_size < $bins);
-
+
-   my ($f_name) = $file =~ /([^\/]+)$/;
-   $f_name =~ s/\.fasta$//;
@@ -538 +546 @@
-   my $f_dir = "$b_dir/$d_name";
-   my $t_dir = $TMP."/$d_name";
-
-
+#
+#
-
-   #delete old indexes if full file does not exist
-
-
-
+#
+#
+#
-
-   #check if files exist
-
-
-
-
-
+
+
+
+
+
+
-
-      my @existing_db;
@@ -558 +567 @@
-      }
-
-
-
-
-
-
-
+
+
+
+
+
+
+
-      }
-      else{ #remove if there is an error
@@ -571 +581 @@
-
-   #make needed output directories
-
+
+
-   mkdir($b_dir) unless (-e $b_dir);
-
-   #open filehandles for  pieces on multi processors
-   my @fhs;
-
+#
-       for (my $i = 0; $i < $bins; $i++) {
-           my $name = "$t_dir/$d_name\.$i\.fasta";
@@ -584 +595 @@
-           push (@fhs, $fh);
-       }
-
+#
-   
-   #write fastas here
-   my %alias;
-
-
+#
+#
-   while (my $fasta = $fasta_iterator->nextEntry()) {
-      my $def = Fasta::getDef(\$fasta);
@@ -606 +617 @@
-      #fix weird blast trimming error for long seq IDs by replacing them
-      if(length($seq_id) > 78){
-substr($seq_id, 0, 78
+uri_escape(Digest::MD5::md5_base64($seq_id), "^A-Za-z0-9\-\_"
-          die "ERROR: The id $seq_id is too long for BLAST, and I can'y uniquely fix it\n"
-              if($alias{$new_id});
-          $alias{$new_id}++;
->/>$seq_id maker_alias=
+(>\S+)/$1 MD5_alias=$new_id
-      }
-
@@ -617 +628 @@
-
-      #build full file
-
+#
-
-      #build part files only on multi processor
-
+#
-          my $fh = shift @fhs;
-          print $fh $$fasta_ref;
-          push (@fhs, $fh);
-
-
-
+#
+
+#
-
-   #close part file handles
@@ -634 +645 @@
-
-   #move finished files into place
-
-
+
+
+
-
-   #check if everything is ok
-
-
-
-
-
+
+
+
+
+
+
-
-      my @db_files;
@@ -649 +662 @@
-      }
-
-
-
-
+
+
+
+
-   }
-   else {
@@ -1029 +1043 @@
-
-         my $fastaObj = $db_index->get_Seq_by_id($hit->name);
+         $fastaObj = $db_index->get_Seq_by_alias($hit->description) if(not $fastaObj);
+
+         #still no sequence? try rebuilding the index and try again
-         if (not $fastaObj) {
-
-
-
-
+
+
-            $fastaObj = $db_index->get_Seq_by_id($hit->name);
+            $fastaObj = $db_index->get_Seq_by_alias($hit->description) if(not $fastaObj);
-            if (not $fastaObj) {
-               print STDERR "stop here:".$hit->name."\n";
@@ -1040 +1056 @@
-            }
-         }
+
-         my $seq      = $fastaObj->seq();
-         my $def      = $db_index->header($hit->name);
+            $def      = $db_index->header_by_alias($hit->description) if(! defined $def);
-
-         my $fasta    = Fasta::toFasta('>'.$def, \$seq);
@@ -1152 +1170 @@
-   foreach my $c (@{$clusters}) {
-      foreach my $hit (@{$c}) {
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-         my $seq      = $fastaObj->seq(); 
-
+
+
-         my $fasta    = Fasta::toFasta('>'.$def, \$seq);
-         $fastas     .= $$fasta; 
@@ -1165 +1197 @@
-sub build_fasta_index {
-   my $db = shift;
-Bio::DB::Fasta
+FastaDB
-   return $index;
-}
@@ -1181 +1213 @@
-   foreach my $db (@dbs){
-       next if(! $db);
-
-
+
+
+
+
+
+
-   }
-}

lib/Widget/RepeatMasker.pm

@@ -9 +9 @@
-use FileHandle;
-use Widget;
-
+#
-use Bio::Search::Hit::PhatHit::repeatmasker;
-use Bio::Search::HSP::PhatHSP::repeatmasker;

lib/Widget/blastn.pm

@@ -99 +99 @@
-         $hit->queryLength($result->query_length);
-         $hit->queryName($result->query_name);
-         if($hit->description() =~ /maker_alias=(\S+)/){
-             $hit->name($1);
-         }
- 
-         my @hsps;

lib/Widget/blastx.pm

@@ -99 +99 @@
-         $hit->queryLength($result->query_length);
-         $hit->queryName($result->query_name);
-
-
-
+
-         my @hsps;
-         while(my $hsp = $hit->next_hsp) {

lib/Widget/tblastx.pm

@@ -99 +99 @@
-         $hit->queryLength($result->query_length);
-         $hit->queryName($result->query_name);
-         if($hit->description() =~ /maker_alias=(\S+)/){
-             $hit->name($1);
-         }
-         
-         my @hsps;