From: elserj <elserj@localhost>
Date: Tue, 16 Dec 2014 23:36:44 +0000 (+0000)
Subject: Add synonym entries
X-Git-Url: http://gitweb.planteome.org/?a=commitdiff_plain;h=48188207522dc1bcd30287af260e79bb9b1d0e82;p=old-jaiswallab-svn%2F.git

Add synonym entries

svn path=/; revision=596
---

diff --git a/interactome_scripts/load_species_db.pl b/interactome_scripts/load_species_db.pl
index 45c5fa8..bdfce67 100755
--- a/interactome_scripts/load_species_db.pl
+++ b/interactome_scripts/load_species_db.pl
@@ -47,14 +47,27 @@ foreach my $file (@files) {
 	
 	my ($retrieval_location, $retrieval_date) = retrieval_info($file);
 	$retrieval_date =~ /(\d+)\_(\d+)\_(\d\d)/;
-	my $formatted_date = "20".$1."-".$2."-".$3;
+	my $formatted_date = "20".$3."-".$1."-".$2;
 	
+	# insert version info into database.  If species previously entered, location and date will be updated.
 	my $safe_info_table = $dbh->quote_identifier("version_info");
-	$dbh->do("insert into $safe_info_table (?,?,?)  on duplicate key update retrieval_location=$retrieval_location,date=$formatted_date",$species,$retrieval_location,$formatted_date);
+	# Create the database for version info if it doesn't already exist
+	$dbh->do("CREATE TABLE IF NOT EXISTS $safe_info_table (
+			`species` varchar(255) NOT NULL,
+			`retrieval_location` varchar(255) NOT NULL,
+			`date` date NOT NULL,
+			PRIMARY KEY (`species`)
+			) ENGINE=MyISAM");
+
+
+	$dbh->do("insert into $safe_info_table (species,retrieval_location,date) values ('$species','$retrieval_location','$formatted_date') on duplicate key update retrieval_location='$retrieval_location', date='$formatted_date'");
 	
-	# create the database table
+	# create the database table for the species
 	my $safe_table = $dbh->quote_identifier($species);
 	
+	# create the database table for all synonyms
+	my $safe_syn_table = $dbh->quote_identifier('synonyms');
+	
 	$dbh->do("drop table if exists $safe_table");
 	$dbh->do("CREATE TABLE $safe_table (
         	`gene_id` VARCHAR( 255 ) NOT NULL ,
@@ -62,16 +75,24 @@ foreach my $file (@files) {
 		`sequence` TEXT NOT NULL ,
 		UNIQUE ( `gene_id` )
 		) TYPE = MYISAM");
-		
+
+	$dbh->do("CREATE TABLE if NOT EXISTS  $safe_syn_table(
+			`gene_id` VARCHAR( 255 ) NOT NULL ,
+			`species` VARCHAR( 255 ) NOT NULL ,
+			`synonyms` TEXT
+			) TYPE = MYISAM");
+
 	# and the statement handler to do the inserts
 	my $insert_sth = $dbh->prepare("insert into $safe_table (gene_id, gene_header, sequence) values (?,?,?)");
+	my $insert_syn_sth = $dbh->prepare("insert into $safe_syn_table (gene_id, species, synonyms) values (?,?,?)");
 	
-	# create the variables:q!
+	# create the variables
 	
 	my %seq_hash; # keys is $gene, values are $seq and $gene_header
 	my $gene;
 	my $seq;
 	my $gene_header;
+	my $synonyms;
 	
 	open(in_file,$file);
 	
@@ -87,6 +108,7 @@ foreach my $file (@files) {
 			$gene_header =~ s/^>//;
 			$gene = find_gene($gene_header,$species);
 			$gene =~ s/^\>//;
+			$synonyms = find_gene_synonym($gene_header,$species);
 			$seq = "";
 		}else{
 			$seq = "$seq"."$line";
@@ -99,6 +121,10 @@ foreach my $file (@files) {
 
 	foreach my $key (keys %seq_hash) {
 		$insert_sth->execute($key,$seq_hash{$key}->{'gene_header'},$seq_hash{$key}->{'sequence'});
+
+		if($synonyms ne "null") {
+				$insert_syn_sth->execute($key,$species,$synonyms);
+		}
 	}
 }