core: add gene panels

opencb · Oct 22, 2018 · 158bdc6 · 158bdc6
1 parent 1d68b15
commit 158bdc6
Show file tree

Hide file tree

Showing 9 changed files with 99 additions and 30 deletions.
diff --git a/bionetdb-app/src/main/java/org/opencb/bionetdb/app/cli/ImportCommandExecutor.java b/bionetdb-app/src/main/java/org/opencb/bionetdb/app/cli/ImportCommandExecutor.java
@@ -96,6 +96,7 @@ private void createCsvFiles() {
 
             long geneIndexingTime = 0;
             long proteinIndexingTime = 0;
+            long genePanelsTime = 0;
             long miRnaIndexingTime = 0;
             long bioPaxTime = 0;
 
@@ -123,6 +124,16 @@ private void createCsvFiles() {
             proteinIndexingTime = (System.currentTimeMillis() - start) / 1000;
             logger.info("Protein indexing done in {} s", proteinIndexingTime);
 
+            // Gene panels support
+            if (Paths.get(inputPath + "/" + Neo4jCsvImporter.PANEL_DIRNAME).toFile().exists()) {
+                logger.info("Starting gene panels processing...");
+                start = System.currentTimeMillis();
+                importer.addGenePanels(Paths.get(inputPath + "/" + Neo4jCsvImporter.PANEL_DIRNAME), outputPath);
+                genePanelsTime = (System.currentTimeMillis() - start) / 1000;
+                logger.info("Gene panels processing done in {} s", genePanelsTime);
+            }
+
+
             // Indexing miRNA
             if (Paths.get(inputPath + "/" + Neo4jCsvImporter.MIRNA_FILENAME).toFile().exists()) {
                 logger.info("Starting miRNA indexing...");
@@ -153,6 +164,7 @@ private void createCsvFiles() {
 
             logger.info("Gene indexing in {} s", geneIndexingTime);
             logger.info("Protein indexing in {} s", proteinIndexingTime);
+            logger.info("Gene panels processing in {} s", genePanelsTime);
             logger.info("miRNA indexing in {} s", miRnaIndexingTime);
             logger.info("BioPAX processing in {} s", bioPaxTime);
             logger.info("Variant processing in {} s", variantTime);

diff --git a/bionetdb-core/pom.xml b/bionetdb-core/pom.xml
@@ -20,6 +20,10 @@
             <groupId>org.opencb.biodata</groupId>
             <artifactId>biodata-formats</artifactId>
         </dependency>
+        <dependency>
+            <groupId>org.opencb.biodata</groupId>
+            <artifactId>biodata-models</artifactId>
+        </dependency>
         <dependency>
             <groupId>org.opencb.commons</groupId>
             <artifactId>commons-lib</artifactId>

diff --git a/bionetdb-core/src/main/java/org/opencb/bionetdb/core/network/Node.java b/bionetdb-core/src/main/java/org/opencb/bionetdb/core/network/Node.java
@@ -85,9 +85,7 @@ public enum Type {
         EXON_OVERLAP("EXON_OVERLAP"),
         PROTEIN_KEYWORD("PROTEIN_KEYWORD"),
 
-        PANEL("PANEL"),
-        DISEASE_GROUP("DISEASE_GROUP"),
-        DISEASE_SUBGROUP("DISEASE_SUBGROUP");
+        PANEL("PANEL");
 
         private final String type;
         private final String parentType;

diff --git a/bionetdb-core/src/main/java/org/opencb/bionetdb/core/network/Relation.java b/bionetdb-core/src/main/java/org/opencb/bionetdb/core/network/Relation.java
@@ -128,11 +128,12 @@ public enum Type {
         PROTEIN__PROTEIN_FEATURE("PROTEIN__PROTEIN_FEATURE"),
         SAMPLE__VARIANT_CALL("SAMPLE__VARIANT_CALL"),
         VARIANT_FILE_INFO__FILE("VARIANT_FILE_INFO__FILE"),
-        VARIANT_CALL__VARIANT_FILE_INFO("VARIANT_CALL__VARIANT_FILE_INFO");
+        VARIANT_CALL__VARIANT_FILE_INFO("VARIANT_CALL__VARIANT_FILE_INFO"),
 
 //        DNA__GENE("DNA__GENE");
 
-//        PANEL__GENE("PANEL__GENE"),
+        PANEL__GENE("PANEL__GENE");
+
 //        DISEASE_GROUP__PANEL("DISEASE_GROUP__PANEL"),
 //        DISEASE_SUBGROUP__PANEL("DISEASE_SUBGROUP__PANEL"),
 //        DISEASE_SUBGROUP__DISEASE_GROUP("DISEASE_SUBGROUP__DISEASE_GROUP"),

diff --git a/bionetdb-core/src/main/java/org/opencb/bionetdb/core/utils/CsvInfo.java b/bionetdb-core/src/main/java/org/opencb/bionetdb/core/utils/CsvInfo.java
@@ -592,6 +592,11 @@ private Map<String, List<String>> createNodeAttributes() {
                 "source", "status");
         nodeAttributes.put(Node.Type.GENE.toString(), new ArrayList<>(attrs));
 
+        // Panel
+        attrs = Arrays.asList("panelId", "id", "name", "description", "phenotypeNames", "sourceId", "sourceName",
+                "sourceAuthor", "sourceProject", "sourceVersion", "creationDate", "modificationDate");
+        nodeAttributes.put(Node.Type.PANEL.toString(), new ArrayList<>(attrs));
+
         // Drug
         attrs = Arrays.asList("drugId", "id", "name", "source", "type", "studyType");
         nodeAttributes.put(Node.Type.DRUG.toString(), new ArrayList<>(attrs));

diff --git a/bionetdb-core/src/main/java/org/opencb/bionetdb/core/utils/Neo4jCsvImporter.java b/bionetdb-core/src/main/java/org/opencb/bionetdb/core/utils/Neo4jCsvImporter.java
@@ -1,11 +1,13 @@
 package org.opencb.bionetdb.core.utils;
 
 import com.fasterxml.jackson.databind.ObjectMapper;
+import com.fasterxml.jackson.databind.ObjectReader;
 import org.apache.commons.lang.StringUtils;
 import org.opencb.biodata.formats.protein.uniprot.v201504jaxb.DbReferenceType;
 import org.opencb.biodata.formats.protein.uniprot.v201504jaxb.Entry;
 import org.opencb.biodata.formats.protein.uniprot.v201504jaxb.FeatureType;
 import org.opencb.biodata.formats.protein.uniprot.v201504jaxb.KeywordType;
+import org.opencb.biodata.models.clinical.interpretation.DiseasePanel;
 import org.opencb.biodata.models.core.Gene;
 import org.opencb.biodata.models.core.Transcript;
 import org.opencb.biodata.models.core.TranscriptTfbs;
@@ -21,17 +23,16 @@
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
-import java.io.BufferedReader;
-import java.io.File;
-import java.io.IOException;
-import java.io.PrintWriter;
+import java.io.*;
 import java.nio.file.Path;
 import java.util.*;
 
 public class Neo4jCsvImporter {
     public static final Object GENE_FILENAME = "genes.json";
     public static final Object GENE_DBNAME = "genes.rocksdb";
 
+    public static final Object PANEL_DIRNAME = "panels";
+
     public static final Object PROTEIN_FILENAME = "proteins.json";
     public static final Object PROTEIN_DBNAME = "proteins.rocksdb";
 
@@ -675,6 +676,43 @@ public void indexingProteins(Path proteinPath, Path indexPath) throws IOExceptio
         csv.indexingProteins(proteinPath, indexPath);
     }
 
+    public void addGenePanels(Path panelPath, Path indexPath) throws IOException {
+        File[] panelFiles = panelPath.toFile().listFiles();
+
+        ObjectMapper mapper = new ObjectMapper();
+        ObjectReader reader = mapper.reader(DiseasePanel.class);
+
+        // Get CSV file writers
+        PrintWriter pwNode = csv.getCsvWriters().get(Node.Type.PANEL.toString());
+        PrintWriter pwRel = csv.getCsvWriters().get(Relation.Type.PANEL__GENE.toString());
+
+        for (File panelFile: panelFiles) {
+            if (panelFile.getName().endsWith("json")) {
+                FileInputStream fis = new FileInputStream(panelFile);
+                byte[] data = new byte[(int) panelFile.length()];
+                fis.read(data);
+                fis.close();
+
+                //String str = new String(data, "UTF-8");
+                DiseasePanel panel = reader.readValue(data);
+
+                // Create node and save CSV file
+                Node node = NodeBuilder.newNode(csv.getAndIncUid(), panel);
+                pwNode.println(csv.nodeLine(node));
+
+                for (DiseasePanel.GenePanel gene: panel.getGenes()) {
+                    if (StringUtils.isNotEmpty(gene.getId())) {
+                        Long geneUid = processGene(gene.getId(), gene.getName());
+                        if (geneUid != null) {
+                            // Add relation to CSV file
+                            pwRel.println(node.getUid() + "," + geneUid);
+                        }
+                    }
+                }
+            }
+        }
+    }
+
     public void indexingMiRnas(Path miRnaPath, Path indexPath, boolean toImport) throws IOException {
         csv.indexingMiRnas(miRnaPath, indexPath);
 

diff --git a/bionetdb-core/src/main/java/org/opencb/bionetdb/core/utils/Neo4jImporter.java b/bionetdb-core/src/main/java/org/opencb/bionetdb/core/utils/Neo4jImporter.java
@@ -314,13 +314,13 @@ public Neo4jImporter() {
         attrs = Arrays.asList("panelId", "id", "name", "author", "version", "date", "sourceProject", "sourceId", "sourceVersion");
         nodeAttributes.put(Node.Type.PANEL.toString(), new ArrayList<>(attrs));
 
-        // Disease group
-        attrs = Arrays.asList("diseaseGroupId", "id", "name");
-        nodeAttributes.put(Node.Type.DISEASE_GROUP.toString(), new ArrayList<>(attrs));
-
-        // Disease subgroup
-        attrs = Arrays.asList("diseaseSubGroupId", "id", "name");
-        nodeAttributes.put(Node.Type.DISEASE_SUBGROUP.toString(), new ArrayList<>(attrs));
+//        // Disease group
+//        attrs = Arrays.asList("diseaseGroupId", "id", "name");
+//        nodeAttributes.put(Node.Type.DISEASE_GROUP.toString(), new ArrayList<>(attrs));
+//
+//        // Disease subgroup
+//        attrs = Arrays.asList("diseaseSubGroupId", "id", "name");
+//        nodeAttributes.put(Node.Type.DISEASE_SUBGROUP.toString(), new ArrayList<>(attrs));
 
         // Ontology
         attrs = Arrays.asList("ontologyId", "id", "name", "source");

diff --git a/bionetdb-core/src/main/java/org/opencb/bionetdb/core/utils/NodeBuilder.java b/bionetdb-core/src/main/java/org/opencb/bionetdb/core/utils/NodeBuilder.java
@@ -2,6 +2,8 @@
 
 import org.apache.commons.lang.StringUtils;
 import org.opencb.biodata.formats.protein.uniprot.v201504jaxb.*;
+import org.opencb.biodata.models.clinical.interpretation.DiseasePanel;
+import org.opencb.biodata.models.commons.Phenotype;
 import org.opencb.biodata.models.core.Gene;
 import org.opencb.biodata.models.core.Transcript;
 import org.opencb.biodata.models.core.TranscriptTfbs;
@@ -299,24 +301,33 @@ public static Node newNode(long uid, DbReferenceType xref) {
         node.addAttribute("dbName", xref.getType());
         return node;
     }
-/*
-    public static Node newNode(long uid, Panel panel) {
+
+    public static Node newNode(long uid, DiseasePanel panel) {
         Node node = new Node(uid, panel.getId(), panel.getName(), Node.Type.PANEL);
-        node.addAttribute("author", panel.getAuthor());
-        node.addAttribute("version", panel.getVersion());
-        node.addAttribute("date", panel.getDate());
+        node.addAttribute("description", panel.getDescription());
+        node.addAttribute("creationDate", panel.getCreationDate());
+        node.addAttribute("modificationDate", panel.getModificationDate());
+        if (ListUtils.isNotEmpty(panel.getPhenotypes())) {
+            StringBuilder sb = new StringBuilder();
+            for (Phenotype phenotype: panel.getPhenotypes()) {
+                if (StringUtils.isNotEmpty(phenotype.getName())) {
+                    if (sb.length() > 0) {
+                        sb.append("--");
+                    }
+                    sb.append(phenotype.getName());
+                }
+            }
+            if (sb.length() > 0) {
+                node.addAttribute("phenotypeNames", sb.toString());
+            }
+        }
         if (panel.getSource() != null) {
-            node.addAttribute("sourceProject", panel.getSource().getProject());
             node.addAttribute("sourceId", panel.getSource().getId());
+            node.addAttribute("sourceName", panel.getSource().getId());
+            node.addAttribute("sourceAuthor", panel.getSource().getAuthor());
+            node.addAttribute("sourceProject", panel.getSource().getProject());
             node.addAttribute("sourceVersion", panel.getSource().getVersion());
         }
         return node;
     }
-
-    public static Node newNode(long uid, OntologyTerm term) {
-        Node node = new Node(uid, term.getId(), term.getName(), Node.Type.ONTOLOGY);
-        node.addAttribute("source", term.getSource());
-        return node;
-    }
-    */
 }
diff --git a/pom.xml b/pom.xml
@@ -23,7 +23,7 @@
         <bionetdb.version>0.2.0-SNAPSHOT</bionetdb.version>
         <java.version>1.8</java.version>
         <java-common-libs.version>3.7.0-SNAPSHOT</java-common-libs.version>
-        <biodata.version>1.4.0-SNAPSHOT</biodata.version>
+        <biodata.version>1.4.2-SNAPSHOT</biodata.version>
         <cellbase.version>4.6.0-SNAPSHOT</cellbase.version>
         <jackson.version>2.8.10</jackson.version>
         <jersey.version>2.23</jersey.version>