From 048352b103d41e8bacf97fdc53cca5ca8c7f1c67 Mon Sep 17 00:00:00 2001
From: steve-fischer-200 <sfischer@upenn.edu>
Date: Mon, 2 Feb 2026 22:02:25 -0500
Subject: [PATCH] update comm dataset with metadata

---
 Model/lib/wdk/Shared/commDatasetQueries.xml | 124 ++++++++++++++++++--
 Model/lib/wdk/Shared/commDatasetRecord.xml  |  95 +++++++++++----
 Model/lib/wdk/Shared/datasetRecord.xml      |  31 ++---
 3 files changed, 192 insertions(+), 58 deletions(-)

diff --git a/Model/lib/wdk/Shared/commDatasetQueries.xml b/Model/lib/wdk/Shared/commDatasetQueries.xml
index e4aeef9..f05d5aa 100644
--- a/Model/lib/wdk/Shared/commDatasetQueries.xml
+++ b/Model/lib/wdk/Shared/commDatasetQueries.xml
@@ -50,12 +50,11 @@
 
     <sqlQuery name="hyperlinks">
       <column name="dataset_id"/>
-      <column name="url"/>
       <column name="text"/>
       <column name="description"/>
       <sql>
         <![CDATA[
-          select aud.user_dataset_id as dataset_id, url, text, description
+          select aud.user_dataset_id as dataset_id, text, description
           from @VDI_CONTROL_SCHEMA@AvailableUserDatasets aud,
 	  @VDI_CONTROL_SCHEMA@dataset_hyperlinks dh	 
           where aud.project_id = '@PROJECT_ID@'
@@ -70,7 +69,8 @@
       <column name="organism_name"/>
       <sql>
         <![CDATA[
-          select aud.user_dataset_id as dataset_id, organism_name
+          select aud.user_dataset_id as dataset_id,
+	  species || ' ' || strain as organism_name
           from @VDI_CONTROL_SCHEMA@AvailableUserDatasets aud,
 	  @VDI_CONTROL_SCHEMA@dataset_organism do, apidbtuning.organismAttributes oa	 
           where aud.project_id = '@PROJECT_ID@'
@@ -88,17 +88,125 @@
       <sql>
         <![CDATA[
           select aud.user_dataset_id as dataset_id,
-          dc.name as contact_name, affiliation
+          concat(dc.first_name, ' ', dc.middle_name, ' ', dc.last_name) as contact_name, affiliation
           from @VDI_CONTROL_SCHEMA@AvailableUserDatasets aud,
-	  @VDI_CONTROL_SCHEMA@dataset_contact dc	 
+	  @VDI_CONTROL_SCHEMA@dataset_contact dc
           where aud.project_id = '@PROJECT_ID@'
           and aud.is_public = true and aud.is_owner = 1
-          and aud.dataset_id = dh.dataset_id
+          and aud.dataset_id = dc.dataset_id
         ]]>
       </sql>
     </sqlQuery>
-    
-    
+
+    <sqlQuery name="countries">
+      <column name="dataset_id"/>
+      <column name="country"/>
+      <sql>
+        <![CDATA[
+          select aud.user_dataset_id as dataset_id, country
+          from @VDI_CONTROL_SCHEMA@AvailableUserDatasets aud,
+	  @VDI_CONTROL_SCHEMA@dataset_country dc
+          where aud.project_id = '@PROJECT_ID@'
+          and aud.is_public = true and aud.is_owner = 1
+          and aud.dataset_id = dc.dataset_id
+        ]]>
+      </sql>
+    </sqlQuery>
+
+    <sqlQuery name="species">
+      <column name="dataset_id"/>
+      <column name="species"/>
+      <sql>
+        <![CDATA[
+          select aud.user_dataset_id as dataset_id, species
+          from @VDI_CONTROL_SCHEMA@AvailableUserDatasets aud,
+	  @VDI_CONTROL_SCHEMA@dataset_species ds
+          where aud.project_id = '@PROJECT_ID@'
+          and aud.is_public = true and aud.is_owner = 1
+          and aud.dataset_id = ds.dataset_id
+        ]]>
+      </sql>
+    </sqlQuery>
+
+    <sqlQuery name="diseases">
+      <column name="dataset_id"/>
+      <column name="disease"/>
+      <sql>
+        <![CDATA[
+          select aud.user_dataset_id as dataset_id, disease
+          from @VDI_CONTROL_SCHEMA@AvailableUserDatasets aud,
+	  @VDI_CONTROL_SCHEMA@dataset_disease dd
+          where aud.project_id = '@PROJECT_ID@'
+          and aud.is_public = true and aud.is_owner = 1
+          and aud.dataset_id = dd.dataset_id
+        ]]>
+      </sql>
+    </sqlQuery>
+
+    <sqlQuery name="associatedFactors">
+      <column name="dataset_id"/>
+      <column name="factor"/>
+      <sql>
+        <![CDATA[
+          select aud.user_dataset_id as dataset_id, factor
+          from @VDI_CONTROL_SCHEMA@AvailableUserDatasets aud,
+	  @VDI_CONTROL_SCHEMA@dataset_associated_factor daf
+          where aud.project_id = '@PROJECT_ID@'
+          and aud.is_public = true and aud.is_owner = 1
+          and aud.dataset_id = daf.dataset_id
+        ]]>
+      </sql>
+    </sqlQuery>
+
+    <sqlQuery name="dois">
+      <column name="dataset_id"/>
+      <column name="doi"/>
+      <column name="description"/>
+      <sql>
+        <![CDATA[
+          select aud.user_dataset_id as dataset_id, doi, description
+          from @VDI_CONTROL_SCHEMA@AvailableUserDatasets aud,
+	  @VDI_CONTROL_SCHEMA@dataset_doi dd
+          where aud.project_id = '@PROJECT_ID@'
+          and aud.is_public = true and aud.is_owner = 1
+          and aud.dataset_id = dd.dataset_id
+        ]]>
+      </sql>
+    </sqlQuery>
+
+    <sqlQuery name="bioprojectIds">
+      <column name="dataset_id"/>
+      <column name="bioproject_id"/>
+      <column name="description"/>
+      <sql>
+        <![CDATA[
+          select aud.user_dataset_id as dataset_id, bioproject_id, description
+          from @VDI_CONTROL_SCHEMA@AvailableUserDatasets aud,
+	  @VDI_CONTROL_SCHEMA@dataset_bioproject_id dbi
+          where aud.project_id = '@PROJECT_ID@'
+          and aud.is_public = true and aud.is_owner = 1
+          and aud.dataset_id = dbi.dataset_id
+        ]]>
+      </sql>
+    </sqlQuery>
+
+    <sqlQuery name="fundingAwards">
+      <column name="dataset_id"/>
+      <column name="agency"/>
+      <column name="award_number"/>
+      <sql>
+        <![CDATA[
+          select aud.user_dataset_id as dataset_id, agency, award_number
+          from @VDI_CONTROL_SCHEMA@AvailableUserDatasets aud,
+	  @VDI_CONTROL_SCHEMA@dataset_funding_award dfa
+          where aud.project_id = '@PROJECT_ID@'
+          and aud.is_public = true and aud.is_owner = 1
+          and aud.dataset_id = dfa.dataset_id
+        ]]>
+      </sql>
+    </sqlQuery>
+
+
   </querySet>
 
 </wdkModel>
diff --git a/Model/lib/wdk/Shared/commDatasetRecord.xml b/Model/lib/wdk/Shared/commDatasetRecord.xml
index a68085a..0792d97 100644
--- a/Model/lib/wdk/Shared/commDatasetRecord.xml
+++ b/Model/lib/wdk/Shared/commDatasetRecord.xml
@@ -1,9 +1,5 @@
 <wdkModel>
 
-  <constant name="datasetRecordName" includeProjects="ClinEpiDB">Study Name</constant>
-  <constant name="datasetRecordName" excludeProjects="ClinEpiDB">Name</constant>
-
-
   <recordClassSet name="communityDatasetRecordClasses">
 
     <!-- ################################################################### -->
@@ -19,16 +15,7 @@
         <columnRef>dataset_id</columnRef>
       </primaryKey>
 
-      <idAttribute name="primary_key" displayName="%%datasetRecordName%%" excludeProjects="ClinEpiDB">
-        <propertyList name="boost">
-          <value>100</value>
-        </propertyList>
-        <text>
-          <![CDATA[ $$display_name$$ ]]>
-        </text>
-      </idAttribute>
-
-      <idAttribute name="primary_key" displayName="Study Name" includeProjects="ClinEpiDB">
+      <idAttribute name="primary_key" displayName="Dataset Name">
         <propertyList name="boost">
           <value>100</value>
         </propertyList>
@@ -42,13 +29,12 @@
                 scopes=""
                 implementation="org.eupathdb.sitesearch.data.model.report.SolrLoaderReporter"/>
 
-      <!-- for ClinEpi and mbio, boost all fields in dataset record, so that Studies generally come out on top of search results -->	    
       <attributeQueryRef ref="CommunityDatasetAttributes.All">
         <columnAttribute name="project" displayName="dontcare" internal="true"/>  <!-- special field for project filtering -->
         <columnAttribute name="hyperlinkName" displayName="dontcare" internal="true"/> <!-- used for display in ss results -->
         <columnAttribute name="display_name" internal="true"/>
         <columnAttribute name="description" displayName="Description">
-          <propertyList name="boost" includeProjects="ClinEpiDB, MicrobiomeDB"> 
+          <propertyList name="boost"> 
             <value>20</value>
           </propertyList>
 	</columnAttribute>
@@ -56,7 +42,7 @@
           <propertyList name="isSummary">
             <value>true</value>
           </propertyList>
-          <propertyList name="boost" includeProjects="ClinEpiDB, MicrobiomeDB">  
+          <propertyList name="boost">  
             <value>20</value>
           </propertyList>
 	</columnAttribute>		
@@ -65,7 +51,7 @@
       <table name="Contacts"
              displayName="Principal Investigator and Collaborators"
              queryRef="DatasetTables.Contacts">
-        <propertyList name="boost" includeProjects="ClinEpiDB, MicrobiomeDB"> 
+        <propertyList name="boost"> 
           <value>20</value>
         </propertyList>   	      
         <columnAttribute name="dataset_id"/>
@@ -76,12 +62,20 @@
       <table name="Publications"
              displayName="Associated publications"
              queryRef="DatasetTables.Publications">
-        <propertyList name="boost" includeProjects="ClinEpiDB, MicrobiomeDB"> 
+        <propertyList name="boost">
           <value>20</value>
-        </propertyList>   	      
+        </propertyList>
         <columnAttribute name="dataset_id"/>
         <columnAttribute name="pmid" displayName="Pubmed ID"/>
-        <columnAttribute name="citation" displayName="Citation"/> 
+        <columnAttribute name="citation" displayName="Citation"/>
+      </table>
+
+      <table name="Hyperlinks"
+             displayName="Related links"
+             queryRef="DatasetTables.Hyperlinks">
+        <columnAttribute name="dataset_id"/>
+        <columnAttribute name="text" displayName="Link text"/>
+        <columnAttribute name="description" displayName="Description"/>
       </table>
 
       <table name="Organisms"
@@ -102,14 +96,63 @@
           <value>AmoebaDB</value>
           <value>UniDB</value>
         </propertyList>
-        <propertyList name="boost" includeProjects="ClinEpiDB, MicrobiomeDB"> 
-          <value>20</value>
-        </propertyList>   	      
         <columnAttribute name="dataset_id"/>
         <columnAttribute name="organism_name"/>
-	      
+
       </table>
-      
+
+      <table name="Countries"
+             displayName="Countries"
+             queryRef="DatasetTables.Countries">
+        <columnAttribute name="dataset_id"/>
+        <columnAttribute name="country" displayName="Country"/>
+      </table>
+
+      <table name="Species"
+             displayName="Species"
+             queryRef="DatasetTables.Species">
+        <columnAttribute name="dataset_id"/>
+        <columnAttribute name="species" displayName="Species"/>
+      </table>
+
+      <table name="Diseases"
+             displayName="Diseases"
+             queryRef="DatasetTables.Diseases">
+        <columnAttribute name="dataset_id"/>
+        <columnAttribute name="disease" displayName="Disease"/>
+      </table>
+
+      <table name="AssociatedFactors"
+             displayName="Associated factors"
+             queryRef="DatasetTables.AssociatedFactors">
+        <columnAttribute name="dataset_id"/>
+        <columnAttribute name="factor" displayName="Factor"/>
+      </table>
+
+      <table name="DOIs"
+             displayName="DOIs"
+             queryRef="DatasetTables.DOIs">
+        <columnAttribute name="dataset_id"/>
+        <columnAttribute name="doi" displayName="DOI"/>
+        <columnAttribute name="description" displayName="Description"/>
+      </table>
+
+      <table name="BioprojectIds"
+             displayName="Bioproject IDs"
+             queryRef="DatasetTables.BioprojectIds">
+        <columnAttribute name="dataset_id"/>
+        <columnAttribute name="bioproject_id" displayName="Bioproject ID"/>
+        <columnAttribute name="description" displayName="Description"/>
+      </table>
+
+      <table name="FundingAwards"
+             displayName="Funding awards"
+             queryRef="DatasetTables.FundingAwards">
+        <columnAttribute name="dataset_id"/>
+        <columnAttribute name="agency" displayName="Agency"/>
+        <columnAttribute name="award_number" displayName="Award number"/>
+      </table>
+
 
     </recordClass>
     
diff --git a/Model/lib/wdk/Shared/datasetRecord.xml b/Model/lib/wdk/Shared/datasetRecord.xml
index ff219df..f112238 100644
--- a/Model/lib/wdk/Shared/datasetRecord.xml
+++ b/Model/lib/wdk/Shared/datasetRecord.xml
@@ -1,9 +1,5 @@
 <wdkModel>
 
-  <constant name="datasetRecordName" includeProjects="ClinEpiDB">Study Name</constant>
-  <constant name="datasetRecordName" excludeProjects="ClinEpiDB">Name</constant>
-
-
   <recordClassSet name="datasetRecordClasses">
 
     <!-- ################################################################### -->
@@ -19,16 +15,7 @@
         <columnRef>dataset_id</columnRef>
       </primaryKey>
 
-      <idAttribute name="primary_key" displayName="%%datasetRecordName%%" excludeProjects="ClinEpiDB">
-        <propertyList name="boost">
-          <value>100</value>
-        </propertyList>
-        <text>
-          <![CDATA[ $$display_name$$ ]]>
-        </text>
-      </idAttribute>
-
-      <idAttribute name="primary_key" displayName="Study Name" includeProjects="ClinEpiDB">
+      <idAttribute name="primary_key" displayName="Dataset Name">
         <propertyList name="boost">
           <value>100</value>
         </propertyList>
@@ -42,18 +29,17 @@
                 scopes=""
                 implementation="org.eupathdb.sitesearch.data.model.report.SolrLoaderReporter"/>
 
-      <!-- for ClinEpi and mbio, boost all fields in dataset record, so that Studies generally come out on top of search results -->	    
       <attributeQueryRef ref="DatasetAttributes.All">
         <columnAttribute name="project" displayName="dontcare" internal="true"/>  <!-- special field for project filtering -->
         <columnAttribute name="hyperlinkName" displayName="dontcare" internal="true"/> <!-- used for display in ss results -->
         <columnAttribute name="display_name" internal="true"/>
         <columnAttribute name="description" displayName="Description">
-          <propertyList name="boost" includeProjects="ClinEpiDB, MicrobiomeDB"> 
+          <propertyList name="boost"> 
             <value>20</value>
           </propertyList>
 	</columnAttribute>
         <columnAttribute name="acknowledgement" displayName="Acknowledgement">
-          <propertyList name="boost" includeProjects="MicrobiomeDB">
+          <propertyList name="boost">
             <value>20</value>
           </propertyList>
 	</columnAttribute>		
@@ -61,12 +47,12 @@
           <propertyList name="isSummary">
             <value>true</value>
           </propertyList>
-          <propertyList name="boost" includeProjects="ClinEpiDB, MicrobiomeDB">  
+          <propertyList name="boost">  
             <value>20</value>
           </propertyList>
 	</columnAttribute>		
         <columnAttribute name="newcategory" displayName="Category">
-          <propertyList name="boost" includeProjects="ClinEpiDB">
+          <propertyList name="boost">
             <value>20</value>
           </propertyList>
 	</columnAttribute>				
@@ -75,7 +61,7 @@
       <table name="Contacts"
              displayName="Principal Investigator and Collaborators"
              queryRef="DatasetTables.Contacts">
-        <propertyList name="boost" includeProjects="ClinEpiDB, MicrobiomeDB"> 
+        <propertyList name="boost"> 
           <value>20</value>
         </propertyList>   	      
         <columnAttribute name="dataset_id"/>
@@ -86,7 +72,7 @@
       <table name="Publications"
              displayName="Associated publications"
              queryRef="DatasetTables.Publications">
-        <propertyList name="boost" includeProjects="ClinEpiDB, MicrobiomeDB"> 
+        <propertyList name="boost"> 
           <value>20</value>
         </propertyList>   	      
         <columnAttribute name="dataset_id"/>
@@ -112,9 +98,6 @@
           <value>AmoebaDB</value>
           <value>UniDB</value>
         </propertyList>
-        <propertyList name="boost" includeProjects="ClinEpiDB, MicrobiomeDB"> 
-          <value>20</value>
-        </propertyList>   	      
         <columnAttribute name="dataset_id"/>
         <columnAttribute name="organism"/>