From d685c2de4e9d4f98c7751901c91a65d5cf1716cc Mon Sep 17 00:00:00 2001
From: athreyab <athreyab@localhost>
Date: Tue, 24 Jan 2012 19:04:36 +0000
Subject: [PATCH] fixed the accession_right_id issue.

svn path=/; revision=273
---
 .../interactions/interactionPathsFromSif.pl   | 37 ++++++++++++-------
 .../interactions/interactionPathsFromTsv.pl   | 28 +++++++++++---
 2 files changed, 46 insertions(+), 19 deletions(-)

diff --git a/Personnel/athreyab/interactions/interactionPathsFromSif.pl b/Personnel/athreyab/interactions/interactionPathsFromSif.pl
index 90406ec..c4cb9ce 100644
--- a/Personnel/athreyab/interactions/interactionPathsFromSif.pl
+++ b/Personnel/athreyab/interactions/interactionPathsFromSif.pl
@@ -11,6 +11,11 @@ require "interactionPath.pl";
 #use constant SITE_ADJECTIVE => ($ENV{'DevelopmentSite'} ? 'Development': 'Live');
 use constant SITE_ADJECTIVE => 'Development';
 
+
+############First various sub routines are defined.############ 
+############Following it the main subroutine code is at the botton of the file############
+
+
 sub showUsage
 {
 	print "\n\nUsage: \n\tperl interactionsPathsFromSif <option> [<fileName>]\n";
@@ -47,12 +52,14 @@ sub importParalogData
 	open(paralog_file, "$fileName") || die "Error: file '$fileName' can not be opened\n";
 	while(<paralog_file>){
 		my $entry = $_;
+		
 		# strip off newline characters
+		chomp $entry;
       $entry =~ s/\r//g;
       $entry =~ s/\n//g;
       # split the columns into 3 separate variables
       my ($obj_l, $obj_r, $score) = split("\t", $entry);
-      # skip if $ident is empty or Nan
+      # skip
       next if(!defined($score));
       next if($score =~ /D/);
       next if($score eq "");
@@ -95,11 +102,12 @@ sub importOrthologData
 	while(<paralog_file>){
 		my $entry = $_;
 		# strip off newline characters
+		chomp $entry;
       $entry =~ s/\r//g;
       $entry =~ s/\n//g;
       # split the columns into 3 separate variables
       my ($obj_l, $obj_r, $score) = split("\t", $entry);
-      # skip if $ident is empty or Nan
+      # skip
       next if(!defined($score));
       next if($score =~ /D/);
       next if($score eq "");
@@ -145,37 +153,40 @@ sub importSifData
 	while(<interactions_file>){
 		my $entry = $_;
 		# strip off newline characters
+		chomp $entry;
 		$entry =~ s/\r//g;
       $entry =~ s/\n//g;
       # split the columns into separate variables
       my ($accession_left,$interaction_type,$accession_right) = split("\t", $entry);
-      # skip if any of the variables are empty
+      
+      # skip if any of the variables are empty      
       next if(isEmpty($accession_left));
 		next if(isEmpty($interaction_type));
 		next if(isEmpty($accession_right));
       
-		#get object_id for object in the first column
+      #get object_id for object in the first column
 		$accession_left_id = getObjectIdFromAccession($accession_left);
+
 		#get interaction_type_id for object in the first column
 		$interaction_type_id = getInteractionTypeForSifFormat($interaction_type);
+
 		#get object_id for object in the second column
-		$accesion_right_id = getObjectIdFromAccession($accession_right);
-		
-		next if(isEmpty($accession_left_id) || isEmpty($accession_right_id) || isEmpty($interaction_type_id));
+		$accession_right_id = getObjectIdFromAccession($accession_right);
 
+		next if(isEmpty($accession_left_id) || isEmpty($accession_right_id) || isEmpty($interaction_type_id));
+		
 		#check if the interaction is already recorded. Write to database only if it doesn't exist already.
 		$interactionId = getInteractionId($accession_left_id,$interaction_type_id,$accession_right_id,$curator_id);
+		print "interaction id is: $interactionId\n";
 		if(isEmpty($interactionId)){
 			$query = "INSERT INTO Interaction(`object_id_left`,`object_id_right`,`interaction_type_id`,`curator_id`)
-			 VALUES('$accession_left_id','$accession_right_id','interaction_type_id','$curatorId')";
-			 executeDbQuery($query);
-			 print "$accession_left_id\t$interaction_type_id\t$accession_right_id\t$curator_id\n";					
+			 VALUES('$accession_left_id','$accession_right_id','$interaction_type_id','$curatorId')";
+			 executeDbQuery($query);					
 		}
 	}
-	print "Seed interaction data import completed";
+	print "Seed interaction data import completed\n";
 }
 
-
 $optionId = $ARGV[0];
 if($optionId ne "0" && $#ARGV != 1){
 	showUsage();
@@ -229,7 +240,7 @@ switch($optionId){
 		print "\tncbi id of the species\n";
 		$curatorId = getCuratorId(getCuratorProperties());
 		$speciesIdForSif = getSpeciesId(getSpeciesProperties(0));	
-		$speciesIdForSif ne "" || die "";			
+		$speciesIdForSif ne "" || die "Species id was not found for ".getSpeciesProperties(0)."\n";			
 		importSifData();
 	}
 		else
diff --git a/Personnel/athreyab/interactions/interactionPathsFromTsv.pl b/Personnel/athreyab/interactions/interactionPathsFromTsv.pl
index 7bb8152..291b4cf 100644
--- a/Personnel/athreyab/interactions/interactionPathsFromTsv.pl
+++ b/Personnel/athreyab/interactions/interactionPathsFromTsv.pl
@@ -5,11 +5,18 @@
 use DBI;
 use Config::IniFiles;
 use Switch;
+
+#functions common to both interactionPathsFromSif and interactionPathsFromTsv are in this perl script file
 require "interactionPath.pl";
 
 #use constant SITE_ADJECTIVE => ($ENV{'DevelopmentSite'} ? 'Development': 'Live');
 use constant SITE_ADJECTIVE => 'Development';
 
+
+############First various sub routines are defined.############ 
+############Following it the main subroutine code is at the botton of the file############
+
+
 sub showUsage(){
 	print "Usage: \n\tperl interactionPathsFromTsv <option> <fileName>\n";
 	print "options:\n\t";
@@ -26,10 +33,13 @@ sub importSeedInteractionData(){
 	while(<interactions_file>){
 		my $entry = $_;
 		# strip off newline characters
+		chomp $entry;
 		$entry =~ s/\r//g;
       $entry =~ s/\n//g;
+      
       # split the columns into separate variables
       my ($obj_abbr_left,$int_type_left,$mode_of_action,$obj_abbr_right,$int_type_right,$interaction_type,$evidence,$evidence_code,$experiment,$comments) = split("\t", $entry);
+      
       #we need all the 3 values before we can insert data into the interaction table.
       next if(isEmpty($obj_abbr_left));
       next if(isEmpty($obj_abbr_right));
@@ -39,6 +49,7 @@ sub importSeedInteractionData(){
       $obj_id_left = getObjectIdFromAbbreviation($obj_abbr_left);
       $obj_id_right = getObjectIdFromAbbreviation($obj_abbr_right);
       $interaction_type_id = getInteractionTypeId($interaction_type);
+     	
      	#if any of the above values are empty, it means they are empty. print that message and move to the next line
 		if(isEmpty($obj_id_left)){
 			print "$obj_abbr_left was not found in the db. Skipping this line\n";
@@ -52,6 +63,8 @@ sub importSeedInteractionData(){
 			print "$interaction_type was not found in the db. Skipping this line\n";
 			next;
 		}
+		
+		#get various primary ids from the table.      
       $int_type_id_left = getInteractorTypeId($int_type_left);
       $int_type_id_right = getInteractorTypeId($int_type_right);
       $mode_of_action_id = getModeOfActionId($mode_of_action);
@@ -59,14 +72,11 @@ sub importSeedInteractionData(){
       $evidence_code_id = getEvidenceCodeId($evidence_code);
       $experiment_id = getExperimentId($experiment);
       
-      print $obj_id_left."\t".$obj_id_right."\t".$int_type_id_left."\t".$int_type_id_right."\t".$interaction_type_id.
-      "\t".$evidence_id."\t".$evidence_code_id."\t".$experiment_id.$curator_id."\n";
-      
+     
       $query = "INSERT INTO Interaction(`object_id_left`,`object_id_right`,`interaction_type_id`,`mode_of_action_id`,
       `interactor_type_id_left`,`interactor_type_id_right`,`evidence_id`,`evidence_code_id`,`experiment_id`,`comments`,`curator_id`) 
       VALUES ('$obj_id_left','$obj_id_right','$interaction_type_id','$mode_of_action_id','$int_type_id_left','$int_type_id_right'
-      ,'$evidence_id','$evidence_code_id','$experiment_id','$comments','$curator_id')";
-      
+      ,'$evidence_id','$evidence_code_id','$experiment_id','$comments','$curator_id')";      
       executeDbQuery($query);
 				
 	}
@@ -77,20 +87,24 @@ sub importGeneInformation(){
 	open(gene_file, "$fileName") || die "Error: file '$fileName' can not be opened\n";
 	while(<gene_file>){
 		my $entry = $_;
+		
 		# strip off newline characters
+		chomp $entry;
 		$entry =~ s/\r//g;
       $entry =~ s/\n//g;
       
       my ($accession, $is_gene,$is_metabolite,$symbol,$full_name,$synonym,$species_name,$ncbi_id) = split("\t", $entry);
+      
       #skip the line if accession value is empty
       next if(isEmpty($accession));
+      
       $object_type=1;
 		if($is_metabolite eq "yes"){
 			$object_type=2;
 			}
 		
+		#get species id, if the species is not present already, insert it
 		$species_id = getSpeciesId($species_name,$ncbi_id);
-		#if the species is not present already, insert it		
 		if(isEmpty($species_id)){
 			if(!isEmpty($ncbi_id)){
 				$query = "INSERT INTO Species(`species`,`NCBI_taxonomy_id`) VALUES ('$species_name','$ncbi_id')";
@@ -100,11 +114,13 @@ sub importGeneInformation(){
 			}
 			executeDbQuery($query);
 			$species_id = getSpeciesId($species_name,$ncbi_id);
+			print "New species $species_name added to the db\n";
 		}
 		
 		#gets synonym_id if exists already. else, inserts synonym and returns the id.
 		$synonym_id = getSynonymId($synonym);
 		
+		#construct the query and insert object information to the table.
 		$query = "INSERT INTO Object(`object_accession`,`species_id`,`object_full_name`,`object_abbreviation`,`object_type_id`) VALUES('$accession','$species_id','$full_name','$symbol','$object_type') ON DUPLICATE KEY UPDATE object_accession='$accession' ";
 		if(!isEmpty($full_name)){
 			$query = $query." , object_full_name='$full_name'";	
-- 
2.34.1