Merge pull request #10 from jdesaphy/patch-biorj_variant

Patch biorj variant

Author: jdesaphy

BACKEND/INSTALL/QUICKSTART/README.md

@@ -52,7 +52,7 @@ While still in $TG_DIR/BACKEND/INSTALL, please execute the following command:
 biorels_php gen_single_script.php > biorels_execute.sh
 sh biorels_execute.sh
 
-This will execute the necessary scripts in the correct order.
+This will execute the necessary scripts in the correct order. It will take about an hour and a half.
 
 
 ## Querying the database
@@ -76,3 +76,4 @@ Please go to $TG_DIR/BACKEND/SCRIPT/BIORJ/
 You can execute the following scripts:
 * biorels_php api_import.php --JSON Example/HeLa.biorj
 * biorels_php api_export.php -JSON_OUTPUT=CVCL_B6YM.biorj --JSON_PRETTY_PRINT CELL CVCL_B6YM && more CVCL_B6YM.biorj
+* biorels_php api_import.php --JSON Example/rs712.json

BACKEND/INSTALL/biorels_public.sql

@@ -3272,14 +3272,14 @@ COMMENT ON COLUMN DB_SCHEMA_NAME.org_group_map.org_group_id IS 'Foreign key to o
 COMMENT ON COLUMN DB_SCHEMA_NAME.org_group_map.web_user_id IS 'Foreign key to web_user representing a scientist';
 
 CREATE SEQUENCE DB_SCHEMA_NAME.org_group_map_sq
-    START WITH 3822421
+    START WITH 1
     INCREMENT BY 1
     NO MINVALUE
     NO MAXVALUE
     CACHE 1;
 
 CREATE SEQUENCE DB_SCHEMA_NAME.org_group_sq
-    START WITH 75516
+    START WITH 1
     INCREMENT BY 1
     NO MINVALUE
     NO MAXVALUE
@@ -5121,7 +5121,7 @@ COMMENT ON COLUMN DB_SCHEMA_NAME.source.source_type IS 'Type of database/resourc
 COMMENT ON COLUMN DB_SCHEMA_NAME.source.source_metadata IS 'Additional information';
 
 CREATE SEQUENCE DB_SCHEMA_NAME.source_seq
-    START WITH 1194
+    START WITH 1
     INCREMENT BY 1
     NO MINVALUE
     NO MAXVALUE
@@ -5420,8 +5420,7 @@ CREATE TABLE DB_SCHEMA_NAME.variant_frequency (
     variant_change_id bigint NOT NULL,
     variant_freq_study_id smallint NOT NULL,
     ref_count integer NOT NULL,
-    alt_count integer NOT NULL,
-    source_id smallint
+    alt_count integer NOT NULL
 );
 
 COMMENT ON TABLE DB_SCHEMA_NAME.variant_frequency IS 'Frequency in different studies of a given allele';
@@ -5436,10 +5435,9 @@ COMMENT ON COLUMN DB_SCHEMA_NAME.variant_frequency.ref_count IS 'Reference count
 
 COMMENT ON COLUMN DB_SCHEMA_NAME.variant_frequency.alt_count IS 'Total count';
 
-COMMENT ON COLUMN DB_SCHEMA_NAME.variant_frequency.source_id IS 'Foreign key to source.';
 
 CREATE SEQUENCE DB_SCHEMA_NAME.variant_frequency_sq
-    START WITH 18778296255
+    START WITH 1
     INCREMENT BY 1
     NO MINVALUE
     NO MAXVALUE
@@ -5659,7 +5657,7 @@ CREATE SEQUENCE DB_SCHEMA_NAME.web_job_document_sq
     CACHE 1;
 
 CREATE SEQUENCE DB_SCHEMA_NAME.web_job_sq
-    START WITH 1640
+    START WITH 1
     INCREMENT BY 1
     NO MINVALUE
     NO MAXVALUE
@@ -5734,7 +5732,7 @@ COMMENT ON COLUMN DB_SCHEMA_NAME.web_user.manager_id IS 'Foreign key to web_user
 COMMENT ON COLUMN DB_SCHEMA_NAME.web_user.worker_type IS 'Type of worker - employee or contingent';
 
 CREATE SEQUENCE DB_SCHEMA_NAME.web_user_seq
-    START WITH 2000
+    START WITH 1
     INCREMENT BY 1
     NO MINVALUE
     NO MAXVALUE
@@ -7390,6 +7388,9 @@ ALTER TABLE ONLY DB_SCHEMA_NAME.variant_freq_study
 ALTER TABLE ONLY DB_SCHEMA_NAME.variant_frequency
     ADD CONSTRAINT variant_frequency_pkey PRIMARY KEY (variant_frequency_id);
 
+ALTER TABLE ONLY DB_SCHEMA_NAME.variant_frequency
+    ADD CONSTRAINT variant_frequency_uk UNIQUE (variant_change_id, variant_freq_study_id);
+
 ALTER TABLE ONLY DB_SCHEMA_NAME.variant_info
     ADD CONSTRAINT variant_info_pkey PRIMARY KEY (variant_info_id);
 

BACKEND/INSTALL/runInstall.php

@@ -4,16 +4,22 @@
 $TG_DIR=getenv('TG_DIR');
 if ($TG_DIR===false)die('TG_DIR not set in setenv.sh');/// TG_DIR is not set as environment variable
 if (!is_dir($TG_DIR))die('Unable to find directory '.$TG_DIR);//// TG_DIR is not an existing directory
-if (!is_file($TG_DIR.'/BACKEND/CONTAINER/env-file.txt'))die('No environment file found');
-if (!is_file($TG_DIR.'/BACKEND/CONTAINER/biorels_container.sif'))echo ('Warning - No container file found');
+if (!is_file($TG_DIR.'/BACKEND/CONTAINER/env-file.txt'))die('No environment file found'."\n");
+if (!is_file($TG_DIR.'/BACKEND/CONTAINER/biorels_container.sif'))echo ('Warning - No container file found'."\n");
+
+$AGREE_ALL=false;
+if (isset($argv[1]) && $argv[1]=='--agree') $AGREE_ALL=true;
+
 
 echo "Root directory for biorels: ".$TG_DIR."\n";
-$resSTDIN=fopen("php://stdin","r");
+if (!$AGREE_ALL)
+{
+	$resSTDIN=fopen("php://stdin","r");
     echo("Do you confirm? Y/N. Then press return: ");
     $strChar = stream_get_contents($resSTDIN, 1);
     if ($strChar=='N') die('You did not agreed'."\n");
     if ($strChar!='Y') die('We didn\'t understood the answer'."\n");
-
+}
 /// Getting environment variables defined in setenv.sh. Here it's DB_SCHEMA, the name of the public schema
 $DB_SCHEMA=getenv('DB_SCHEMA');	if ($DB_SCHEMA===false) die('No schema provided');
 /// Getting environment variables defined in setenv.sh. Here it's PRIVATE_SCHEMA, the name of the private schema
@@ -22,12 +28,14 @@
 
 
 echo "\n\n\nPublic schema:".$DB_SCHEMA."\n";
+if (!$AGREE_ALL)
+{
 	$resSTDIN=fopen("php://stdin","r");
     echo("This will delete this schema if it exists. Do you confirm? Y/N. Then press return: ");
     $strChar = stream_get_contents($resSTDIN, 1);
     if ($strChar=='N') die('You did not agreed'."\n");
     if ($strChar!='Y') die('We didn\'t understood the answer'."\n");
-
+}
 
 echo "\n\nCreating tables for ".$DB_SCHEMA."\n";
 /// Here we are going to convert the sql template file into the installation file, which includes the correct schema name.
@@ -47,11 +55,14 @@
 if ($PRIVATE_SCHEMA!==false)
 {
 	echo "\n\n\n\nPrivate schema:".$PRIVATE_SCHEMA."\n";
+	if (!$AGREE_ALL)
+	{
 	$resSTDIN=fopen("php://stdin","r");
     echo("This will delete this schema if it exists. Do you confirm? Y/N. Then press return: ");
     $strChar = stream_get_contents($resSTDIN, 1);
     if ($strChar=='N') die('You did not agreed'."\n");
     if ($strChar!='Y') die('We didn\'t understood the answer'."\n");
+	}
 	$change=array('DB_SCHEMA_NAME'=>$DB_SCHEMA,'DB_PRIVATE_SCHEMA'=>$PRIVATE_SCHEMA);
 //// Here we convert the template sql file by changing the template schema name to the private schema name
 	convertFile($TG_DIR.'/BACKEND/INSTALL/biorels_private.sql',$TG_DIR.'/BACKEND/INSTALL/private_schema_ready.sql',$change);

BACKEND/SCRIPT/API/queries.php

@@ -2774,14 +2774,12 @@ function get_translation_isoform_pos($ISOFORM_ID,$POSITION)
 # // Title: Search gene ontology record by name and/or accession and/or namespace
 # // Function: search_gene_ontology
 # // Description: Search for a gene ontology record by using its name and/or accession and/or namespace
-# // Parameter: AC | Gene ontology accession | string | GO:0010389 | optional
+# // Parameter: AC | Gene ontology accession | string | GO:0010389 
 # // Parameter: NAME | Gene ontology name | string | cell cycle | optional
 # // Parameter: NAMESPACE | Gene ontology namespace | string | biological_process |  optional
 # // Return: Gene ontology record
 # // Ecosystem: Proteomics:gene ontology
 # // Example: php biorels_api.php search_gene_ontology -AC 'GO:0010389'
-# // Example: php biorels_api.php search_gene_ontology -NAME 'cell cycle'
-# // Example: php biorels_api.php search_gene_ontology -NAMESPACE 'biological_process'
 # // $[/API]
 function search_gene_ontology($AC='',$NAME='',$NAMESPACE='')
 {

BACKEND/SCRIPT/BIORJ/BIORJ_RULES

@@ -155,14 +155,15 @@ CHILD	drug_atc_map<atc_entry
 #CHILD	drug_mol_entity_map
 END
 
-BLOCK	clinical_variant
+BLOCK	clinical_variant_entry
 PARENT	clinical_variant_review_status
-CHILD	clinical_variant_disease_map<disease_entry
-CHILD	clinical_variant_gn_map<gn_entry
-CHILD	clinical_variant_map<variant_entry
-CHILD	clinical_variant_pmid_map<pmid_entry:E
+PARENT	clinical_variant_type<so_entry
+CHILD	clinical_variant_map<variant_entry:E
+CHILD	clinical_variant_submission>clinical_variant_pmid_map<pmid_entry:E
 CHILD	clinical_variant_submission<clinical_variant_review_status
 CHILD	clinical_variant_submission<clinical_significance
+CHILD	clinical_variant_submission>clinical_variant_disease_map<disease_entry:E
+CHILD	clinical_variant_submission>clinical_variant_gn_map<gn_entry:E
 END
 
 
@@ -195,6 +196,7 @@ BLOCK	variant_entry
 CHILD	variant_position<chr_seq_pos<chr_seq<chromosome<taxon
 CHILD	variant_position<chr_seq_pos<chr_seq<genome_assembly<taxon
 CHILD	variant_position<variant_allele
+CHILD	clinical_variant_map<clinical_variant_entry:E
 CHILD	variant_position>variant_change<variant_type<so_entry
 CHILD	variant_position>variant_change<variant_allele
 CHILD	variant_position>variant_change>variant_frequency<variant_freq_study<source
@@ -306,6 +308,7 @@ clinical_trial_intervention	intervention_name|clinical_trial_id|intervention_typ
 clinical_trial_intervention_drug_map	clinical_trial_intervention_id|drug_entry_id|source_id
 clinical_trial_arm_intervention_map	clinical_trial_arm_id|clinical_trial_intervention_id
 clinical_trial_alias	alias_name|clinical_trial_id
+clinical_variant_entry	clinical_variant_name
 clinical_variant_disease_map	disease_entry_id|clinvar_submission_id
 clinical_variant_gn_map	gn_entry_id|clinvar_submission_id
 clinical_variant_map	clinvar_entry_id|variant_entry_id
@@ -382,7 +385,7 @@ pmid_abstract	pmid_entry_id|abstract_type
 pmid_citation	pmid_entry_id|citation_pmid_entry_id
 pmid_company_map	company_entry_id|pmid_entry_id
 pmid_instit	instit_hash
-pmid_journal	journal_name
+pmid_journal	journal_name|issn_print
 prot_ac	ac|prot_entry_id
 prot_desc	prot_entry_id|desc_type|description
 prot_desc_pmid	prot_desc_id|pmid_entry_id

BACKEND/SCRIPT/BIORJ/Example/rs712.json

@@ -13,6 +13,8 @@
 	'/LIB/loader_process.php'=>0
 );
 
+$DEBUG=false;
+
 function microtime_float()
 {
     list($usec, $sec) = explode(" ", microtime());

BACKEND/SCRIPT/BIORJ/api_export.php

@@ -49,6 +49,7 @@ function microtime_float()
 
 connectDB();
 
+$DEBUG=false;
 
 
 if ($argc<2)