Set an object's default version to 1 when it is created, rather than when it...

Set an object's default version to 1 when it is created, rather than when it is stored, to allow undef versions (which can then be stored as NULLs in the database).

modules/Bio/EnsEMBL/DBSQL/ExonAdaptor.pm

@@ -383,9 +383,8 @@ sub store {
 
   if ( defined($exon->stable_id) ) {
 
-     $exonst->bind_param( 9, $exon->stable_id, SQL_VARCHAR );
-     my $version = ($exon->version()) ? $exon->version() : 1;
-     $exonst->bind_param( 10, $version, SQL_INTEGER ); 
+     $exonst->bind_param(  9, $exon->stable_id, SQL_VARCHAR );
+     $exonst->bind_param( 10, $exon->version,   SQL_INTEGER ); 
   }
 
   $exonst->execute();

modules/Bio/EnsEMBL/DBSQL/GeneAdaptor.pm

@@ -1340,8 +1340,7 @@ sub store {
   if (defined($gene->stable_id)) {
 
     $sth->bind_param(12, $gene->stable_id, SQL_VARCHAR);
-    my $version = ($gene->version()) ? $gene->version() : 1;
-    $sth->bind_param(13, $version, SQL_INTEGER);
+    $sth->bind_param(13, $gene->version,   SQL_INTEGER);
   }
 
   $sth->execute();

modules/Bio/EnsEMBL/DBSQL/TranscriptAdaptor.pm

@@ -1174,8 +1174,7 @@ sub store {
   if ( defined( $transcript->stable_id() ) ) {
 
     $tst->bind_param( ++$i, $transcript->stable_id(), SQL_VARCHAR );
-    my $version = ($transcript->version()) ? $transcript->version() : 1;
-    $tst->bind_param( ++$i, $version,                 SQL_INTEGER );
+    $tst->bind_param( ++$i, $transcript->version(),   SQL_INTEGER );
   }
 
   $tst->execute();

modules/Bio/EnsEMBL/DBSQL/TranslationAdaptor.pm

@@ -522,8 +522,7 @@ sub store {
   if (defined($translation->stable_id)) {
  
     $sth->bind_param(6, $translation->stable_id,SQL_VARCHAR);
-    my $version = ($translation->version()) ? $translation->version() : 1;
-    $sth->bind_param(7, $version,SQL_VARCHAR);
+    $sth->bind_param(7, $translation->version,SQL_VARCHAR);
   }
 
   $sth->execute();

modules/Bio/EnsEMBL/Exon.pm

@@ -138,10 +138,13 @@ sub new {
 
   $self->{'end_phase'}     = $end_phase;
   $self->{'stable_id'}     = $stable_id;
-  $self->{'version'}       = $version;
   $self->{'created_date'}  = $created_date;
   $self->{'modified_date'} = $modified_date;
 
+  # Default version
+  if ( !defined($version) ) { $version = 1 }
+  $self->{'version'} = $version;
+
   # Default is_current
   if ( !defined($is_current) ) { $is_current = 1 }
   $self->{'is_current'} = $is_current;

modules/Bio/EnsEMBL/Gene.pm

@@ -165,7 +165,6 @@ sub new {
   }
 
   $self->stable_id($stable_id);
-  $self->version($version);
   $self->{'created_date'}  = $created_date;
   $self->{'modified_date'} = $modified_date;
 
@@ -182,6 +181,10 @@ sub new {
   $self->status($status);    # add new naming
   $self->source($source);
 
+  # Default version
+  if ( !defined($version) ) { $version = 1 }
+  $self->{'version'} = $version;
+
   # default to is_current
   $is_current = 1 unless (defined($is_current));
   $self->{'is_current'} = $is_current;

modules/Bio/EnsEMBL/Transcript.pm

@@ -172,7 +172,6 @@ sub new {
   }
 
   $self->stable_id($stable_id);
-  $self->version($version);
   $self->{'created_date'}  = $created_date;
   $self->{'modified_date'} = $modified_date;
   $self->external_name($external_name) if ( defined $external_name );
@@ -188,6 +187,10 @@ sub new {
   $self->biotype($biotype);
   $self->source($source);
 
+  # Default version
+  if ( !defined($version) ) { $version = 1 }
+  $self->{'version'} = $version;
+
   # default is_current
   $is_current = 1 unless ( defined($is_current) );
   $self->{'is_current'} = $is_current;

modules/Bio/EnsEMBL/Translation.pm

@@ -113,6 +113,9 @@ sub new {
 	   rearrange( [ "START_EXON", "END_EXON", "SEQ_START", "SEQ_END",
 			"STABLE_ID", "VERSION", "DBID", "ADAPTOR",
 			"SEQ", "CREATED_DATE", "MODIFIED_DATE" ], @_ );
+      
+  # Default version
+  if ( !defined($version) ) { $version = 1 }
 
   my $self = bless {
 		    'start_exon' => $start_exon,

modules/t/exon.t

@@ -281,8 +281,7 @@ my $e1 = Bio::EnsEMBL::Exon->new(
   -slice => $slice,
   -phase => 0,
   -end_phase => 0,
-  -stable_id => 'ENSE0001',
-  -version => 1
+  -stable_id => 'ENSE0001'
 );
 
 my $e2 = Bio::EnsEMBL::Exon->new(
@@ -297,17 +296,40 @@ my $e2 = Bio::EnsEMBL::Exon->new(
   -is_current => 0
 );
 
+my $e3 = Bio::EnsEMBL::Exon->new(
+  -start => 10,
+  -end => 1000,
+  -strand => 1,
+  -slice => $slice,
+  -phase => 0,
+  -end_phase => 0,
+  -stable_id => 'ENSE0001',
+  -is_current => 0
+);
+$e3->version(undef);
+
 $exonad->store($e1);
 $exonad->store($e2);
+$exonad->store($e3);
 
 $exon = $exonad->fetch_by_stable_id('ENSE0001');
 ok( $exon->is_current == 1);
+ok( $exon->version == 1);
 
 @exons = @{ $exonad->fetch_all_versions_by_stable_id('ENSE0001') };
 foreach my $e (@exons) {
-  next unless ($e->version == 2);
-  ok($e->is_current == 0);
+  if (defined $e->version && $e->version == 2) {
+    ok($e->is_current == 0);
+  }
+}
+
+my $null_versions = 0;
+foreach my $e (@exons) {
+  if (! defined $e->version) {
+    $null_versions++;
+  }
 }
+is ( $null_versions, 1, "Null/undef version stored and retrieved");
 
 $multi->restore();
 

modules/t/gene.t

@@ -140,6 +140,11 @@ my $translation2 = Bio::EnsEMBL::Translation->new();
 
 ok($gene);
 
+is($gene->version,         1, 'Default gene version = 1');
+is($transcript1->version,  1, 'Default transcript version = 1');
+is($ex1->version,          1, 'Default exon version = 1');
+is($translation1->version, 1, 'Default translation version = 1');
+
 $ex1->start(13586);
 $ex1->end(13735);
 $ex1->phase(0);
@@ -324,6 +329,9 @@ my $gene_out = $genes->[0];
 #make sure the stable_id was stored
 ok($gene_out->stable_id eq $stable_id);
 
+#make sure the version was stored
+ok($gene_out->version == 1);
+
 #make sure the description was stored
 ok($gene_out->description eq $desc);
 
@@ -923,13 +931,20 @@ $gene->dbID(undef);
 $gene->adaptor(undef);
 $ga->store($gene);
 
+$gene->version(undef);
+$gene->is_current(0);
+$gene->dbID(undef);
+$gene->adaptor(undef);
+$ga->store($gene);
+
 $gene = $ga->fetch_by_stable_id('ENSG00000355555');
 ok($gene->is_current == 1);
 
 @genes = @{$ga->fetch_all_versions_by_stable_id('ENSG00000355555')};
 foreach my $g (@genes) {
-  next unless ($g->version == 4);
-  ok($g->is_current == 0);
+  if (defined $g->version && $g->version == 4) {
+    ok($g->is_current == 0);
+  }
 }
 
 $gene->is_current(0);
@@ -942,6 +957,13 @@ $ga->update($gene);
 $gene = $ga->fetch_by_stable_id('ENSG00000355555');
 ok($gene->is_current == 1);
 
+my $null_versions = 0;
+foreach my $g (@genes) {
+  if (! defined $g->version) {
+    $null_versions++;
+  }
+}
+is ( $null_versions, 1, "Null/undef version stored and retrieved");
 
 $ga->remove_by_Slice($slice);
 $geneCount = $ga->count_all_by_Slice($slice);

modules/t/gtfSerialiser.t

@@ -71,15 +71,15 @@ my $db = $mtdb->get_DBAdaptor("core");
   my $gtf_serializer = Bio::EnsEMBL::Utils::IO::GTFSerializer->new($fh);
   $gtf_serializer->print_Gene($gene);
   my $gtf = <<GTF;
-20\tensembl\tgene\t30274331\t30274404\t.\t+\t.\tgene_id \"GENE\"; gene_source \"ensembl\"; gene_biotype \"protein_coding\"; projection_parent_gene \"ENSG_PARENT_GENE\";
-20\tensembl\ttranscript\t30274331\t30274404\t.\t+\t.\tgene_id \"GENE\"; transcript_id \"TRANS\"; gene_source \"ensembl\"; gene_biotype \"protein_coding\"; projection_parent_gene \"ENSG_PARENT_GENE\"; transcript_source \"ensembl\"; transcript_biotype \"protein_coding\"; tag \"CCDS\"; ccds_id \"CCDS.1\"; tag \"seleno\"; projection_parent_transcript "ENST_PARENT_TRANSCRIPT";
-20\tensembl\tSelenocysteine\t30274337\t30274339\t.\t+\t.\tgene_id \"GENE\"; transcript_id \"TRANS\"; gene_source \"ensembl\"; gene_biotype \"protein_coding\"; projection_parent_gene \"ENSG_PARENT_GENE\"; transcript_source \"ensembl\"; transcript_biotype \"protein_coding\"; tag \"CCDS\"; ccds_id \"CCDS.1\"; tag \"seleno\"; projection_parent_transcript "ENST_PARENT_TRANSCRIPT";
-20\tensembl\texon\t30274331\t30274348\t.\t+\t.\tgene_id \"GENE\"; transcript_id \"TRANS\"; exon_number \"1\"; gene_source \"ensembl\"; gene_biotype \"protein_coding\"; projection_parent_gene \"ENSG_PARENT_GENE\"; transcript_source \"ensembl\"; transcript_biotype \"protein_coding\"; tag \"CCDS\"; ccds_id \"CCDS.1\"; exon_id \"e1\"; tag \"seleno\"; projection_parent_transcript "ENST_PARENT_TRANSCRIPT";
-20\tensembl\tCDS\t30274334\t30274345\t.\t+\t0\tgene_id \"GENE\"; transcript_id \"TRANS\"; exon_number \"1\"; gene_source \"ensembl\"; gene_biotype \"protein_coding\"; projection_parent_gene \"ENSG_PARENT_GENE\"; transcript_source \"ensembl\"; transcript_biotype \"protein_coding\"; tag \"CCDS\"; ccds_id \"CCDS.1\"; protein_id \"PEP\"; tag \"seleno\"; projection_parent_transcript "ENST_PARENT_TRANSCRIPT";
-20\tensembl\texon\t30274401\t30274404\t.\t+\t.\tgene_id \"GENE\"; transcript_id \"TRANS\"; exon_number \"2\"; gene_source \"ensembl\"; gene_biotype \"protein_coding\"; projection_parent_gene \"ENSG_PARENT_GENE\"; transcript_source \"ensembl\"; transcript_biotype \"protein_coding\"; tag \"CCDS\"; ccds_id \"CCDS.1\"; exon_id \"e2\"; tag \"seleno\"; projection_parent_transcript "ENST_PARENT_TRANSCRIPT";
-20\tensembl\tfive_prime_utr\t30274331\t30274333\t.\t+\t.\tgene_id \"GENE\"; transcript_id \"TRANS\"; gene_source \"ensembl\"; gene_biotype \"protein_coding\"; projection_parent_gene \"ENSG_PARENT_GENE\"; transcript_source \"ensembl\"; transcript_biotype \"protein_coding\"; tag \"CCDS\"; ccds_id \"CCDS.1\"; tag \"seleno\"; projection_parent_transcript "ENST_PARENT_TRANSCRIPT";
-20\tensembl\tthree_prime_utr\t30274346\t30274348\t.\t+\t.\tgene_id \"GENE\"; transcript_id \"TRANS\"; gene_source \"ensembl\"; gene_biotype \"protein_coding\"; projection_parent_gene \"ENSG_PARENT_GENE\"; transcript_source \"ensembl\"; transcript_biotype \"protein_coding\"; tag \"CCDS\"; ccds_id \"CCDS.1\"; tag \"seleno\"; projection_parent_transcript "ENST_PARENT_TRANSCRIPT";
-20\tensembl\tthree_prime_utr\t30274401\t30274404\t.\t+\t.\tgene_id \"GENE\"; transcript_id \"TRANS\"; gene_source \"ensembl\"; gene_biotype \"protein_coding\"; projection_parent_gene \"ENSG_PARENT_GENE\"; transcript_source \"ensembl\"; transcript_biotype \"protein_coding\"; tag \"CCDS\"; ccds_id \"CCDS.1\"; tag \"seleno\"; projection_parent_transcript "ENST_PARENT_TRANSCRIPT";
+20\tensembl\tgene\t30274331\t30274404\t.\t+\t.\tgene_id \"GENE\"; gene_version \"1\"; gene_source \"ensembl\"; gene_biotype \"protein_coding\"; projection_parent_gene \"ENSG_PARENT_GENE\";
+20\tensembl\ttranscript\t30274331\t30274404\t.\t+\t.\tgene_id \"GENE\"; gene_version \"1\"; transcript_id \"TRANS\"; transcript_version \"1\"; gene_source \"ensembl\"; gene_biotype \"protein_coding\"; projection_parent_gene \"ENSG_PARENT_GENE\"; transcript_source \"ensembl\"; transcript_biotype \"protein_coding\"; tag \"CCDS\"; ccds_id \"CCDS.1\"; tag \"seleno\"; projection_parent_transcript "ENST_PARENT_TRANSCRIPT";
+20\tensembl\tSelenocysteine\t30274337\t30274339\t.\t+\t.\tgene_id \"GENE\"; gene_version \"1\"; transcript_id \"TRANS\"; transcript_version \"1\"; gene_source \"ensembl\"; gene_biotype \"protein_coding\"; projection_parent_gene \"ENSG_PARENT_GENE\"; transcript_source \"ensembl\"; transcript_biotype \"protein_coding\"; tag \"CCDS\"; ccds_id \"CCDS.1\"; tag \"seleno\"; projection_parent_transcript "ENST_PARENT_TRANSCRIPT";
+20\tensembl\texon\t30274331\t30274348\t.\t+\t.\tgene_id \"GENE\"; gene_version \"1\"; transcript_id \"TRANS\"; transcript_version \"1\"; exon_number \"1\"; gene_source \"ensembl\"; gene_biotype \"protein_coding\"; projection_parent_gene \"ENSG_PARENT_GENE\"; transcript_source \"ensembl\"; transcript_biotype \"protein_coding\"; tag \"CCDS\"; ccds_id \"CCDS.1\"; exon_id \"e1\"; exon_version \"1\"; tag \"seleno\"; projection_parent_transcript "ENST_PARENT_TRANSCRIPT";
+20\tensembl\tCDS\t30274334\t30274345\t.\t+\t0\tgene_id \"GENE\"; gene_version \"1\"; transcript_id \"TRANS\"; transcript_version \"1\"; exon_number \"1\"; gene_source \"ensembl\"; gene_biotype \"protein_coding\"; projection_parent_gene \"ENSG_PARENT_GENE\"; transcript_source \"ensembl\"; transcript_biotype \"protein_coding\"; tag \"CCDS\"; ccds_id \"CCDS.1\"; protein_id \"PEP\"; protein_version \"1\"; tag \"seleno\"; projection_parent_transcript "ENST_PARENT_TRANSCRIPT";
+20\tensembl\texon\t30274401\t30274404\t.\t+\t.\tgene_id \"GENE\"; gene_version \"1\"; transcript_id \"TRANS\"; transcript_version \"1\"; exon_number \"2\"; gene_source \"ensembl\"; gene_biotype \"protein_coding\"; projection_parent_gene \"ENSG_PARENT_GENE\"; transcript_source \"ensembl\"; transcript_biotype \"protein_coding\"; tag \"CCDS\"; ccds_id \"CCDS.1\"; exon_id \"e2\"; exon_version \"1\"; tag \"seleno\"; projection_parent_transcript "ENST_PARENT_TRANSCRIPT";
+20\tensembl\tfive_prime_utr\t30274331\t30274333\t.\t+\t.\tgene_id \"GENE\"; gene_version \"1\"; transcript_id \"TRANS\"; transcript_version \"1\"; gene_source \"ensembl\"; gene_biotype \"protein_coding\"; projection_parent_gene \"ENSG_PARENT_GENE\"; transcript_source \"ensembl\"; transcript_biotype \"protein_coding\"; tag \"CCDS\"; ccds_id \"CCDS.1\"; tag \"seleno\"; projection_parent_transcript "ENST_PARENT_TRANSCRIPT";
+20\tensembl\tthree_prime_utr\t30274346\t30274348\t.\t+\t.\tgene_id \"GENE\"; gene_version \"1\"; transcript_id \"TRANS\"; transcript_version \"1\"; gene_source \"ensembl\"; gene_biotype \"protein_coding\"; projection_parent_gene \"ENSG_PARENT_GENE\"; transcript_source \"ensembl\"; transcript_biotype \"protein_coding\"; tag \"CCDS\"; ccds_id \"CCDS.1\"; tag \"seleno\"; projection_parent_transcript "ENST_PARENT_TRANSCRIPT";
+20\tensembl\tthree_prime_utr\t30274401\t30274404\t.\t+\t.\tgene_id \"GENE\"; gene_version \"1\"; transcript_id \"TRANS\"; transcript_version \"1\"; gene_source \"ensembl\"; gene_biotype \"protein_coding\"; projection_parent_gene \"ENSG_PARENT_GENE\"; transcript_source \"ensembl\"; transcript_biotype \"protein_coding\"; tag \"CCDS\"; ccds_id \"CCDS.1\"; tag \"seleno\"; projection_parent_transcript "ENST_PARENT_TRANSCRIPT";
 GTF
   eq_or_diff(${$fh->string_ref}, $gtf, 'Checking custom Gene object dumps UTRs, Selenocysteine, seleno tag and CCDS');
 }

modules/t/transcript.t

@@ -729,8 +729,9 @@ $tr = $ta->fetch_by_stable_id('ENST00000355555');
 $g = $db->get_GeneAdaptor->fetch_by_transcript_id($tr->dbID);
 $tr->get_all_Exons;
 
-$multi->hide( "core", "gene", "transcript", "exon", 'xref', 'object_xref',
-              "exon_transcript", "translation", 'meta_coord' );
+my $tl = $tr->translation;
+
+$multi->hide( "core", "gene", "transcript", "translation", "meta_coord" );
 
 $tr->version(3);
 $tr->dbID(undef);
@@ -742,13 +743,23 @@ $tr->is_current(0);
 $tr->dbID(undef);
 $tr->adaptor(undef);
 $ta->store($tr, $g->dbID);
+
+$tr->version(undef);
+$tr->is_current(0);
+$tr->dbID(undef);
+$tr->adaptor(undef);
+$tl->version(undef);
+$tr->translation($tl);
+$ta->store($tr, $g->dbID);
+
 $tr = $ta->fetch_by_stable_id('ENST00000355555');
 is($tr->is_current, 1, 'Transcript is current');   # 148
 
 @transcripts = @{ $ta->fetch_all_versions_by_stable_id('ENST00000355555') };
 foreach my $t (@transcripts) {
-  next unless ($t->version == 4);
-  is($t->is_current, 0, 'Transcript is not current');  # 149
+  if (defined $t->version && $t->version == 4) {
+    is($t->is_current, 0, 'Transcript is not current');  # 149
+  }
 }
 
 $tr->is_current(0);
@@ -761,6 +772,15 @@ $ta->update($tr);
 $tr = $ta->fetch_by_stable_id('ENST00000355555');
 is($tr->is_current, 1, 'Transcript is now current');   # 151
 
+my $null_versions = 0;
+foreach my $t (@transcripts) {
+  if (! defined $t->version) {
+    ok(! defined $t->translation->version);
+    $null_versions++;
+  }
+}
+is ( $null_versions, 1, "Null/undef version stored and retrieved");
+
 $multi->restore;
 
 # UTR Tests

modules/t/translation.t

@@ -52,6 +52,8 @@ ok(test_getter_setter($t, 'dbID', 3));
 ok(test_getter_setter($t, 'start', 42));
 ok(test_getter_setter($t, 'end',   50));
 
+is($t->version, 1, 'Default translation version = 1');
+
 my $exon = Bio::EnsEMBL::Exon->new();
 $exon->start(10);
 $exon->end(20);