t.display_name bug and new pathways data

yeastgenome · Jan 13, 2020 · 143a2bc · 143a2bc
1 parent b42bb35
commit 143a2bc
Show file tree

Hide file tree

Showing 4 changed files with 56 additions and 26 deletions.
diff --git a/sgd/src/main/java/org/intermine/bio/dataconversion/SgdConverter.java b/sgd/src/main/java/org/intermine/bio/dataconversion/SgdConverter.java
@@ -72,7 +72,7 @@ public class SgdConverter extends BioDBConverter {
 	private static final String TAXON_ID = "4932";
 	private Item organism;
 	private Map<String, String> featureMap = new HashMap();
-	private static final boolean TEST_LOCAL = true;
+	private static final boolean TEST_LOCAL = false;
 	private String licence;
 
 
@@ -135,12 +135,11 @@ public void process() throws Exception {
 		processRegulation(connection);
 		processRegulationSummary(connection);
 
+		processPathways(connection);
+		storePathways();
 
 		if(TEST_LOCAL) {
 
-			processPathways(connection);
-			storePathways();
-
 			processPhysicalInteractions(connection);
 			processGeneticInteractions(connection);
 			storeInteractionTypes();
@@ -181,7 +180,6 @@ private void processGenes(Connection connection) throws SQLException,
 				String symbol = res.getString("gene_name");
 				String name = res.getString("name_description");
 				String feature_type = res.getString("feature_type");
-				System.out.println("feature TYpe got from SQL is " + feature_type);
 				String headline = res.getString("headline");
 				String description = res.getString("description");
 				String qualifier = res.getString("qualifier");
@@ -250,7 +248,6 @@ private void processGenes(Connection connection) throws SQLException,
 
 				// set for all types, so you can use LSF to query for these
 				// different type of objects in a template.
-				System.out.println("feature Type is...."+ feature_type);
 				item.setAttribute("featureType", feature_type);
 				item.setAttribute("primaryIdentifier", primaryIdentifier);
 				if (StringUtils.isNotEmpty(name)) item.setAttribute("name", name);
@@ -608,10 +605,11 @@ private void processPathways(Connection connection)
 			String dbxref_name = res.getString("display_name"); //pathway identifier i.e. short name
 			String summary_type = res.getString("summary_type");
 			String text = res.getString("text");
+			String refNo = res.getString("reference_id"); // refs for pathways
 
 			Item item = genes.get(geneFeatureNo);
 			if (item != null) {
-				getPathway(item.getIdentifier(), dbxref_id, dbxref_name, summary_type, text);
+				getPathway(item.getIdentifier(), dbxref_id, dbxref_name, summary_type, text, refNo);
 			}
 		}
 	}
@@ -1927,17 +1925,18 @@ private void processAllPubs(Connection connection) throws SQLException,
 			String referenceNo = res.getString("dbentity_id");
 			String pubMedId = res.getString("pmid");
 			String title = res.getString("title");
+            String volume = res.getString("volume");
+            String pages = res.getString("page");
+            String year = res.getString("year");
+            String issue = res.getString("issue");
 			String citation = res.getString("citation");
 			String topic = res.getString("topic");
 			String journal = res.getString("med_abbr");
-			String volume = res.getString("volume");
-			String pages = res.getString("page");
-			String year = res.getString("year");
-			String issue = res.getString("issue");
-			String abst = res.getString("fulltext_status");
+			String abstractText = res.getString("text");
 			String dbxrefid = res.getString("sgdid");
 			String date_created = res.getString("date_created");
 
+			System.out.println(abstractText);
 			if (firstrow) {
 				prevReferenceNo = referenceNo;
 				firstrow = false;
@@ -1963,11 +1962,12 @@ private void processAllPubs(Connection connection) throws SQLException,
 			prevYear = year;
 			prevIssue = issue;
 			prevPages = pages;
-			prevAbst = abst;
+			prevAbst = abstractText;
 			prevDbxRef = dbxrefid;
 			prevDateCreated = date_created;
 
 		}
+
 		// process the very last reference group
 		getPub(prevReferenceNo, prevTitle, prevPubMedId, prevCitation, hm,
 				prevJournal, prevVolume, prevPages, prevYear, prevIssue,
@@ -2699,7 +2699,7 @@ private String getSynonym(String subjectId, String type, String value)
 	}
 
 
-	private String getPathway(String geneIdentifier, String id, String name, String summaryType, String text)
+	private String getPathway(String geneIdentifier, String id, String name, String summaryType, String text, String refNo)
 			throws ObjectStoreException {
 
 		Item crf = pathways.get(id);
@@ -2709,13 +2709,26 @@ private String getPathway(String geneIdentifier, String id, String name, String
 			crf.setAttribute("identifier", id);
 			crf.setAttribute("name", name);
 			crf.setAttribute("summaryType", summaryType);
-			crf.setAttribute("text", text);
+			if (!StringUtils.isEmpty(text)) { crf.setAttribute("text", text);}
+
+			Item publication = publications.get(refNo);
+			if(publication == null) {
+				publication = createItem("Publication");
+				publications.put(refNo, publication);
+			}
+			crf.addToCollection("publications", publication);
 
 			pathways.put(id, crf);
 
 			crf.addToCollection("genes", geneIdentifier);
 
 		}else{
+			Item publication = publications.get(refNo);
+			if(publication == null) {
+				publication = createItem("Publication");
+				publications.put(refNo, publication);
+			}
+			crf.addToCollection("publications", publication);
 			crf.addToCollection("genes", geneIdentifier);
 		}
 
@@ -2879,7 +2892,7 @@ private String getLiteratureTopic(String topic) throws ObjectStoreException {
 
 	private void getPub(String referenceNo, String title, String pubMedId,
 			String citation, ArrayList hm, String journal, String volume,
-			String pages, String year, String issue, String abst, String dbxrefid, String datecreated)
+			String pages, String year, String issue, String abstractText, String dbxrefid, String datecreated)
 					throws ObjectStoreException {
 
 		Item storedRef = publications.get(referenceNo);
@@ -2915,8 +2928,8 @@ private void getPub(String referenceNo, String title, String pubMedId,
 			if (StringUtils.isNotEmpty(issue)) {
 				item.setAttribute("issue", issue);
 			}
-			if (StringUtils.isNotEmpty(abst)) {
-				item.setAttribute("summary", abst);
+			if (StringUtils.isNotEmpty(abstractText)) {
+				item.setAttribute("abstractText", abstractText);
 			}
 			if (StringUtils.isNotEmpty(datecreated)) {
 				item.setAttribute("dateCreated", datecreated);

diff --git a/sgd/src/main/java/org/intermine/bio/dataconversion/SgdProcessor.java b/sgd/src/main/java/org/intermine/bio/dataconversion/SgdProcessor.java
@@ -126,7 +126,8 @@ protected ResultSet getChildrenFeatureLocationResults(Connection connection)
 				+ " and    a2.dbentity_id = d2.dbentity_id "
 				+ " and    d2.dbentity_id = ld.dbentity_id "
 				+ " and    a.taxonomy_id = t.taxonomy_id "
-				+ " and    t.display_name = 'Saccharomyces cerevisiae S288c'";
+				+ " and    t.taxid = 'TAX:559292'";
+				//+ " and    t.display_name = 'Saccharomyces cerevisiae S288c'";
 
 		LOG.info("executing: " + query);
 		Statement stmt = connection.createStatement();
@@ -338,7 +339,7 @@ protected ResultSet getPubWithFeaturesResults(Connection connection)
 	protected ResultSet getPubAllResults(Connection connection)
 			throws SQLException {
 
-		String query = "select r.dbentity_id, r.pmid, r.fulltext_status, r.title, r.volume, r.page, r.year, r.issue,"
+		String query = "select r.dbentity_id, r.pmid, r.title, r.volume, r.page, r.year, r.issue,"
 				+	" r.citation, la.topic, j.med_abbr, rd.text, db.sgdid, db.date_created"
 				+ " from nex.dbentity db"
 				+ " inner join nex.referencedbentity r on db.dbentity_id = r.dbentity_id"
@@ -647,13 +648,21 @@ protected ResultSet getCrossReferences(Connection connection)
 	protected ResultSet getPathways(Connection connection)
 			throws SQLException {
 
-		String query = "select distinct db.dbentity_id, biocyc_id, db2.display_name, ps.summary_type, ps.text "
+		/*String query = "select distinct db.dbentity_id, biocyc_id, db2.display_name, ps.summary_type, ps.text "
 				+ " from nex.dbentity db, nex.pathwaydbentity pdf, nex.pathwayannotation pa, nex.dbentity db2, nex.pathwaysummary ps "
 				+ " where db.dbentity_id = pa.dbentity_id "
 				+ " and ps.pathway_id = pdf.dbentity_id "
 				+ " and pa.pathway_id = pdf.dbentity_id "
 				+ " and pdf.dbentity_id = db2.dbentity_id " 
-				+ " order by 1";
+				+ " order by 1";*/
+
+		String query = "select distinct db.dbentity_id, biocyc_id, db2.display_name, ps.summary_type, ps.text, pa.reference_id" +
+				" from nex.dbentity db" +
+				" inner join nex.pathwayannotation pa on db.dbentity_id = pa.dbentity_id" +
+				" inner join nex.pathwaydbentity pdf on pdf.dbentity_id = pa.pathway_id" +
+				" inner join nex.dbentity db2 on db2.dbentity_id = pdf.dbentity_id" +
+				" left join nex.pathwaysummary ps on ps.pathway_id = pdf.dbentity_id" +
+				" order by 1";
 
 		LOG.info("executing: " + query);        
 		Statement stmt = connection.createStatement();

diff --git a/...l-expression/src/main/java/org/intermine/bio/dataconversion/SpellExpressionConverter.java b/...l-expression/src/main/java/org/intermine/bio/dataconversion/SpellExpressionConverter.java
@@ -143,7 +143,7 @@ private void processDataSets(Connection connection) throws SQLException,
 			Item dataset = getDataSet(pubmedID, fileName, geoID, platformID,
 					channelCount, datasetName, description, numCond, numGenes,
 					author, allAuthors, title, journal, pubYear, tags);		
-			System.out.println("gene: "+geneName + "    condDesc: "+ condDesc + " data_table: " + dataTable);
+			//System.out.println("gene: "+geneName + "    condDesc: "+ condDesc + " data_table: " + dataTable);
 
 			// add score to gene - using the condition and dataset info
 			getConditionScore(dataset, condDesc, dataTable, geneName);
@@ -262,7 +262,7 @@ private Item getDataSet(String pubmedID, String fileName, String geoID,
 			         item.addToCollection("expressiondatasettags", dtag.getIdentifier());
 
 			}
-			System.out.println("pubmed: "+pubmedID + "    geoID: "+ geoID + " datasetName: " + datasetName);								
+			//System.out.println("pubmed: "+pubmedID + "    geoID: "+ geoID + " datasetName: " + datasetName);
 			datasets.put(fileName, item);
 		}
 
@@ -290,7 +290,6 @@ private void getConditionScore(Item dataSet, String condDesc,
 
 			String cond = expconditions[i];
 			String condscore = scores[i];
-			System.out.println("cond:" + cond + "   score:"+ condscore);
 
 			Item dcond = conditions.get(cond);
 			if (dcond == null) {

diff --git a/yeastmine-global_additions.xml b/yeastmine-global_additions.xml
@@ -453,11 +453,18 @@
 <class name="Pathway" is-interface="true">
 	<attribute name="identifier" type="java.lang.String"/>
 	<attribute name="summaryType" type="java.lang.String"/>
-	<attribute name="name" type="java.lang.String"/>
 	<attribute name="text" type="java.lang.String"/>
+	<attribute name="name" type="java.lang.String"/>
+	<!--<reference name="summary" referenced-type="PathwaySummary"/>-->
 	<collection name="genes" referenced-type="Gene" reverse-reference="pathways"/>
 	<collection name="dataSets" referenced-type="DataSet"/>
+	<collection name="publications" referenced-type="Publication" reverse-reference="pathways"/>
 </class>
+<!--<class name="PathwaySummary" is-interface="true">
+	<attribute name="summaryType" type="java.lang.String"/>
+	<attribute name="text" type="java.lang.String"/>
+	<collection name="publications" referenced-type="Publication" reverse-reference="pathwaysummary"/>
+</class>-->
 <class name="Phenotype" is-interface="true">
 	<attribute name="mutantType" type="java.lang.String"/>
 	<attribute name="strainBackground" type="java.lang.String"/>
@@ -615,6 +622,8 @@
 	<collection name="literatureTopics" referenced-type="LiteratureTopic"/>
 	<collection name="crossReferences" referenced-type="DatabaseReference" reverse-reference="subject"/>
 	<collection name="meshTerms" referenced-type="MeshTerm" reverse-reference="publications"/>
+	<collection name="pathways" referenced-type="Pathway" reverse-reference="publications"/>
+	<!--<collection name="pathwaysummary" referenced-type="Phenotype" reverse-reference="publications"/>-->
 </class>
 <class name="PublicationAnnotation" is-interface="true">
 	<reference name="publication" referenced-type="Publication"/>