Merge branch 'master' of https://github.com/WormBase/wormbase-pipeline

autoace_config/elegans.config

@@ -70,6 +70,7 @@ db=citace	file=caltech_CDS.ace				class=CDS	format="Interaction WBInteraction\d{
 db=citace	file=caltech_Cell.ace			class=Cell	format="Reference WBPaper\d{8}"	format="Anatomy_term WBbt:\d{7}"
 db=citace	file=caltech_Cell_group.ace		class=Cell_group	format="Reference WBPaper\d{8}"	format="Anatomy_term WBbt:\d{7}"
 db=citace	file=caltech_Condition.ace				class=Condition
+db=citace	file=caltech_Analysis.ace		class=Analysis	query=(Sample)	format="Reference WBPaper\d{8}"	format="Conducted_by WBPerson\d{1,5}"
 db=citace	file=caltech_DO_defs.ace		class=DO_term	format="Reference WBPaper\d{8}"
 db=citace	file=caltech_Database.ace		class=Database
 db=citace	file=caltech_Expr_pattern.ace	class=Expr_pattern	format="Gene WBGene\d{8}"	format="Reference WBPaper\d{8}"

scripts/ENSEMBL/lib/WormBase2Ensembl.pm

@@ -457,12 +457,6 @@ sub parse_genes_gff3_fh {
           my $bt = "ncRNA";
           $transcript->biotype($bt);
           $gene_biotypes{$bt}++;
-        }  elsif ( $gff_type eq 'antisense_RNA') {
-          # not acknowledged as a biotype by Ensembl; change to default ncRNA
-          $transcript->analysis($nc_ana);
-          my $bt = "antisense";
-          $transcript->biotype($bt);
-          $gene_biotypes{$bt}++;
         } else {
           $transcript->analysis($nc_ana);
           my $bt = ($gff_type =~ /RNA/) ? $gff_type : 'ncRNA';

scripts/NAMEDB/geneace_nameDB_comm.pl

@@ -100,8 +100,13 @@
 
 # any genes left in the acedb list are absent from the nameserver
 foreach (keys %ace_genes ){
+    if ($_ =~ /ENS/) {
+	print "Skipping ENSEMBL ID:$_\n" if $debug;
+    }
+    else {
 	$log->error("ERROR: $_ absent from nameserver\n");
 	$errorcount++;
+    }
 }
 $log->write_to("INFO: $errorcount errors found\n") if ($log->report_errors > 0);
 $log->write_to("No errors found\n") if ($log->report_errors == 0);

scripts/NAMEDB/variation_server_compare.pl

@@ -36,9 +36,13 @@
 my $acedb = ($database or $wormbase->database('geneace'));
 $log->write_to("Checking $acedb for errors\n");
 
-my $def = "$ENV{CVS_DIR}/../wquery/geneace/variation_nameserver_comm.def";
+#my $def = "$ENV{CVS_DIR}/../wquery/geneace/variation_nameserver_comm.def";
+my $tmdef = &get_table_maker_def();
+my $command = "Table-maker -p $tmdef\nquit\n";
+$log->write_to("\nRetrieving Variation data, using Table-maker and query ${tmdef}...\n");
 
-my $TABLE = $wormbase->table_maker_query($acedb, $def);
+
+my $TABLE = $wormbase->table_maker_query($acedb, $tmdef);
 my %ace_ids;
 while(<$TABLE>) {
   next unless /WBVar/;
@@ -127,3 +131,46 @@
 $log->write_to("Work Done!\n");
 $log->mail();
 exit(0);
+
+
+sub get_table_maker_def {
+
+  my $def = '/tmp/nsvar.def';
+  open TMP,">$def" or $log->log_and_die("cant write $def: $!\n");
+  my $txt = <<END;
+
+Sortcolumn 1
+
+Colonne 1 
+Width 12 
+Optional 
+Visible 
+Class 
+Class Variation 
+From 1 
+ 
+Colonne 2 
+Width 12 
+Mandatory
+Visible 
+Next_Tag 
+From 1 
+Tag Status
+ 
+Colonne 3 
+Width 12 
+Optional
+Visible 
+Class 
+Class Variation_name 
+From 1 
+Tag Public_name  
+
+// End of these definitions
+END
+
+  print TMP $txt;
+  close TMP;
+
+  return $def;
+}

scripts/check_class.pl

@@ -601,14 +601,10 @@ sub set_classes {
     @classes = (
 		"Sequence",
 		"CDS", 
-		"Transposon",
 		"Transcript",
 		"Pseudogene",
 		"Transposon_CDS",
 		"cDNA_sequence",
-		"${species}_CDS",
-		"${species}_pseudogenes",
-		"${species}_RNA_genes",
 		"Class",
 		"Model",
 		"Method",
@@ -628,8 +624,6 @@ sub set_classes {
 		"Peptide",
 		"Protein",
 		"Species",
-		"Transposon_family",
-		"Comment",
 		"Database",
 		"Display",
 		"DNA",

scripts/check_predicted_genes.pl

@@ -262,6 +262,20 @@
 	$seen{$gg}=1;
       }
 
+      my @isoformer_genes = $db->fetch (-query => "FIND $qclass where \"*iso*\"");
+      foreach my $g (@isoformer_genes) {
+my $gg=$g->name; 
+	if ($ignore{$gg}) {next}
+	if (exists $seen{$gg}) {
+	  $s=' (seen already)';
+	} else {
+	  $s='';
+	}
+	$log->write_to("Error: $qclass $gg should be removed $s\n"); 
+	$seen{$gg}=1;
+      }
+
+
       my @no_Sparent_genes = $db->fetch (-query => "FIND $qclass where !S_parent");
       foreach my $g (@no_Sparent_genes) {
 	my $gg=$g->name; 

scripts/confirm_genes.pl

@@ -1,3 +1,6 @@
+
+
+
 #!/usr/local/bin/perl5.8.0 -w
 #
 # confirm_genes.pl
@@ -620,7 +623,7 @@ sub create_transcript_file {
 	}else {
 		$prefix = $wormbase->chromosome_prefix."${chrom}_";
 	}
-        my $trinity = -e "${prefix}BLAT_Trinity_BEST.gff" ? "${prefix}BLAT_Trinity_BEST.gff" : '';
+        my $trinity = -e "${gffdir}/${prefix}BLAT_Trinity_BEST.gff" ? "${prefix}BLAT_Trinity_BEST.gff" : '';
 	$wormbase->run_command("cd $gffdir; cat ${prefix}BLAT_EST_BEST.gff ${prefix}BLAT_mRNA_BEST.gff $trinity >  ${prefix}BLAT_TRANSCRIPT_BEST.gff", $log) unless (-e "$gffdir/${prefix}BLAT_TRANSCRIPT_BEST.gff");
 
 	# only elegans has OSTs and RSTs

scripts/data_checks.pl

@@ -258,7 +258,7 @@ sub read_GFF_queries {
     $i++;
     $queries[$i]{'DESC'}  = "Allele";
     $queries[$i]{'GFF'}   = "\tAllele\tsequence_alteration";
-    $queries[$i]{'QUERY'} = 'find Variation flanking_sequences AND method = "Allele"';
+    $queries[$i]{'QUERY'} = 'find Variation Flanking_sequences AND (method = "Allele" OR method = "Engineered_allele")';
 
     $i++;
     $queries[$i]{'DESC'}  = "Transposon_insertion";