Merge branch 'TASK-5564' into TASK-5387

opencb · May 7, 2024 · 7972070 · 7972070
2 parents 4ea3ffe + babb593
commit 7972070
Show file tree

Hide file tree

Showing 7 changed files with 261 additions and 1 deletion.
diff --git a/biodata-formats/src/main/java/org/opencb/biodata/formats/feature/mirbase/MirBaseParser.java b/biodata-formats/src/main/java/org/opencb/biodata/formats/feature/mirbase/MirBaseParser.java
@@ -0,0 +1,118 @@
+package org.opencb.biodata.formats.feature.mirbase;
+
+import org.opencb.biodata.models.core.MiRnaGene;
+import org.opencb.biodata.models.core.MiRnaMature;
+import org.opencb.commons.utils.FileUtils;
+
+import java.io.BufferedReader;
+import java.io.IOException;
+import java.io.InputStreamReader;
+import java.nio.file.Path;
+
+public class MirBaseParser {
+
+    private static final String ID_LABEL = "ID";
+    private static final String AC_LABEL = "AC";
+    private static final String DE_LABEL = "DE";
+    private static final String FT_LABEL = "FT";
+    private static final String SQ_LABEL = "SQ";
+    private static final String END_OF_ITEM_LABEL = "XX";
+    private static final String END_OF_RECORD_LABEL = "//";
+
+    private static final String MIRNA_LABEL = "miRNA";
+
+    private MirBaseParser() {
+        throw new IllegalStateException("Utility class");
+    }
+
+    public static void parse(Path miRnaDatFile, String species, MirBaseParserCallback callback) throws IOException {
+        try (BufferedReader datReader = new BufferedReader(new InputStreamReader(FileUtils.newInputStream(miRnaDatFile)))) {
+            String miRBaseAccession = null;
+            String miRBaseID = null;
+            MiRnaGene miRnaGene = null;
+            String line;
+            while ((line = datReader.readLine()) != null) {
+                String[] split = line.split("\\s+");
+                switch (split[0]) {
+                    case ID_LABEL: {
+                        miRBaseID = split[1];
+                        break;
+                    }
+                    case AC_LABEL: {
+                        miRBaseAccession = split[1].split(";")[0];
+                        break;
+                    }
+                    case DE_LABEL: {
+                        if (line.contains(species)) {
+                            miRnaGene = new MiRnaGene();
+                            miRnaGene.setId(miRBaseID)
+                                    .setAccession(miRBaseAccession);
+                        }
+                        break;
+                    }
+                    case FT_LABEL: {
+                        if (miRnaGene != null && MIRNA_LABEL.equalsIgnoreCase(split[1])) {
+                            processMiRnaMature(line, miRnaGene, datReader);
+                        }
+                        break;
+                    }
+                    case SQ_LABEL: {
+                        if (miRnaGene != null) {
+                            StringBuilder seq = new StringBuilder();
+                            // Read until END_OF_RECORD_LABEL
+                            while (!(line = datReader.readLine()).equals(END_OF_RECORD_LABEL)) {
+                                split = line.split("\\s+");
+                                for (int i = 1; i < split.length - 1; i++) {
+                                    seq.append(split[i]);
+                                }
+                            }
+                            miRnaGene.setSequence(seq.toString());
+
+                            // Update mature sequences
+                            for (MiRnaMature mature : miRnaGene.getMatures()) {
+                                if (mature.getStart() > 0 && mature.getEnd() > 0) {
+                                    mature.setSequence(miRnaGene.getSequence().substring(mature.getStart() - 1, mature.getEnd()));
+                                }
+                            }
+
+                            // Callback
+                            callback.processMiRnaGene(miRnaGene);
+                            miRnaGene = null;
+                        }
+                        break;
+                    }
+                    default: {
+                        // Do nothing
+                        break;
+                    }
+                }
+            }
+        }
+    }
+
+    private static void processMiRnaMature(String headerLine, MiRnaGene miRnaGene, BufferedReader datReader) throws IOException {
+        // Create MiRNA mature from header line,
+        // e.g: FT   miRNA           6..27
+        MiRnaMature miRnaMature = new MiRnaMature();
+        String[] split = headerLine.split("\\s+");
+        String[] pos = split[2].split("\\.\\.");
+        miRnaMature.setStart(Integer.parseInt(pos[0]));
+        miRnaMature.setEnd(Integer.parseInt(pos[1]));
+
+        String line;
+        while (!(line = datReader.readLine()).equals(END_OF_ITEM_LABEL)) {
+            split = line.split("\\s+");
+            if (split[0].equalsIgnoreCase(FT_LABEL) && split[1].equalsIgnoreCase(MIRNA_LABEL)) {
+                processMiRnaMature(line, miRnaGene, datReader);
+                break;
+            } else {
+                if (line.contains("accession=")) {
+                    miRnaMature.setAccession(line.split("accession=")[1].replace("\"", ""));
+                } else if (line.contains("product=")) {
+                    miRnaMature.setId(line.split("product=")[1].replace("\"", ""));
+                }
+            }
+        }
+        miRnaGene.getMatures().add(miRnaMature);
+    }
+}
diff --git a/...rmats/src/main/java/org/opencb/biodata/formats/feature/mirbase/MirBaseParserCallback.java b/...rmats/src/main/java/org/opencb/biodata/formats/feature/mirbase/MirBaseParserCallback.java
@@ -0,0 +1,7 @@
+package org.opencb.biodata.formats.feature.mirbase;
+
+import org.opencb.biodata.models.core.MiRnaGene;
+
+public interface MirBaseParserCallback {
+        boolean processMiRnaGene(MiRnaGene miRnaGene);
+}
diff --git a/...a-formats/src/test/java/org/opencb/biodata/formats/feature/mirbase/MirBaseParserTest.java b/...a-formats/src/test/java/org/opencb/biodata/formats/feature/mirbase/MirBaseParserTest.java
@@ -0,0 +1,105 @@
+package org.opencb.biodata.formats.feature.mirbase;
+
+import org.junit.Assert;
+import org.junit.Test;
+import org.opencb.biodata.models.core.MiRnaGene;
+import org.opencb.biodata.models.core.MiRnaMature;
+
+import java.io.IOException;
+import java.nio.file.Path;
+import java.nio.file.Paths;
+import java.util.ArrayList;
+import java.util.List;
+
+public class MirBaseParserTest {
+
+
+
+    // Implementation of the MirBaseParserCallback function
+    public class MyCallback implements MirBaseParserCallback {
+        private String msg;
+        private List<MiRnaGene> miRnaGenes;
+
+        public MyCallback(String msg) {
+            this.msg = msg;
+            this.miRnaGenes = new ArrayList<>();
+        }
+
+        @Override
+        public boolean processMiRnaGene(MiRnaGene miRnaGene) {
+            System.out.println(msg);
+            System.out.println(miRnaGene.toString());
+            miRnaGenes.add(miRnaGene);
+            return true;
+        }
+
+        public List<MiRnaGene> getMiRnaGenes() {
+            return miRnaGenes;
+        }
+
+        public MiRnaGene getMiRnaGene(String accession) {
+            for (MiRnaGene miRnaGene : miRnaGenes) {
+                if (accession.equals(miRnaGene.getAccession())) {
+                    return miRnaGene;
+                }
+            }
+            return null;
+        }
+
+        public int getCounter() {
+            return miRnaGenes.size();
+        }
+    }
+
+    @Test
+    public void testMirBaseParser() throws IOException {
+        Path datFile = Paths.get(getClass().getResource("/miRNA.small.dat.gz").getPath());
+
+        MyCallback callback = new MyCallback(">>> Testing message");
+
+        MirBaseParser.parse(datFile, "Homo sapiens", callback);
+        Assert.assertEquals(50, callback.getCounter());
+
+        MiRnaGene mi0000060 = callback.getMiRnaGene("MI0000060");
+        Assert.assertEquals("hsa-let-7a-1", mi0000060.getId());
+        Assert.assertEquals("ugggaUGAGGUAGUAGGUUGUAUAGUUuuagggucacacccaccacugggagauaaCUAUACAAUCUACUGUCUUUCcua".toUpperCase(), mi0000060.getSequence().toUpperCase());
+        int found = 0;
+        for (MiRnaMature mature : mi0000060.getMatures()) {
+            if ("MIMAT0000062".equals(mature.getAccession())) {
+                found++;
+                Assert.assertEquals("hsa-let-7a-5p", mature.getId());
+                Assert.assertEquals("UGAGGUAGUAGGUUGUAUAGUU".toUpperCase(), mature.getSequence().toUpperCase());
+                Assert.assertEquals(6, mature.getStart());
+                Assert.assertEquals(27, mature.getEnd());
+            } else if ("MIMAT0004481".equals(mature.getAccession())) {
+                found++;
+                Assert.assertEquals("hsa-let-7a-3p", mature.getId());
+                Assert.assertEquals("CUAUACAAUCUACUGUCUUUC".toUpperCase(), mature.getSequence().toUpperCase());
+                Assert.assertEquals(57, mature.getStart());
+                Assert.assertEquals(77, mature.getEnd());
+            }
+        }
+        Assert.assertEquals(2, found);
+
+        MiRnaGene mi0000077 = callback.getMiRnaGene("MI0000077");
+        Assert.assertEquals("hsa-mir-21", mi0000077.getId());
+        Assert.assertEquals("ugucgggUAGCUUAUCAGACUGAUGUUGAcuguugaaucucauggCAACACCAGUCGAUGGGCUGUcugaca".toUpperCase(), mi0000077.getSequence().toUpperCase());
+        found = 0;
+        for (MiRnaMature mature : mi0000077.getMatures()) {
+            if ("MIMAT0000076".equals(mature.getAccession())) {
+                found++;
+                Assert.assertEquals("hsa-miR-21-5p", mature.getId());
+                Assert.assertEquals("UAGCUUAUCAGACUGAUGUUGA".toUpperCase(), mature.getSequence().toUpperCase());
+                Assert.assertEquals(8, mature.getStart());
+                Assert.assertEquals(29, mature.getEnd());
+            } else if ("MIMAT0004494".equals(mature.getAccession())) {
+                found++;
+                Assert.assertEquals("hsa-miR-21-3p", mature.getId());
+                Assert.assertEquals("CAACACCAGUCGAUGGGCUGU".toUpperCase(), mature.getSequence().toUpperCase());
+                Assert.assertEquals(46, mature.getStart());
+                Assert.assertEquals(66, mature.getEnd());
+            }
+        }
+        Assert.assertEquals(2, found);
+    }
+}
diff --git a/biodata-formats/src/test/resources/miRNA.small.dat.gz b/biodata-formats/src/test/resources/miRNA.small.dat.gz
diff --git a/biodata-models/src/main/java/org/opencb/biodata/models/core/MiRnaGene.java b/biodata-models/src/main/java/org/opencb/biodata/models/core/MiRnaGene.java
@@ -40,7 +40,7 @@ public class MiRnaGene {
 	private List<MiRnaMature> matures;
 
 	public MiRnaGene() {
-
+		matures = new ArrayList<>();
 	}
 
 	public MiRnaGene(String accession, String id, String status, String sequence, List<MiRnaMature> matures) {

diff --git a/biodata-models/src/main/java/org/opencb/biodata/models/pharma/guideline/Literature.java b/biodata-models/src/main/java/org/opencb/biodata/models/pharma/guideline/Literature.java
@@ -8,7 +8,9 @@ public class Literature {
     private String _sameAs;
     private List<CrossReference> crossReferences;
     private String objCls;
+    private String pubDate;
     private List<Term> terms;
+    private String type;
 
     public float getId() {
         return id;
@@ -55,6 +57,15 @@ public Literature setObjCls(String objCls) {
         return this;
     }
 
+    public String getPubDate() {
+        return pubDate;
+    }
+
+    public Literature setPubDate(String pubDate) {
+        this.pubDate = pubDate;
+        return this;
+    }
+
     public List<Term> getTerms() {
         return terms;
     }
@@ -63,6 +74,15 @@ public Literature setTerms(List<Term> terms) {
         this.terms = terms;
         return this;
     }
+
+    public String getType() {
+        return type;
+    }
+
+    public Literature setType(String type) {
+        this.type = type;
+        return this;
+    }
 }
 
 
diff --git a/...odels/src/main/java/org/opencb/biodata/models/pharma/guideline/PharmaDosingGuideline.java b/...odels/src/main/java/org/opencb/biodata/models/pharma/guideline/PharmaDosingGuideline.java
@@ -15,6 +15,7 @@ public class PharmaDosingGuideline {
     private boolean hasTestingInfo;
     private List<History> history;
     private List<Literature> literature;
+    private boolean otherPrescribingGuidance;
     private boolean pediatric;
     private PediatricMarkdown pediatricMarkdown;
     private boolean recommendation;
@@ -136,6 +137,15 @@ public PharmaDosingGuideline setLiterature(List<Literature> literature) {
         return this;
     }
 
+    public boolean isOtherPrescribingGuidance() {
+        return otherPrescribingGuidance;
+    }
+
+    public PharmaDosingGuideline setOtherPrescribingGuidance(boolean otherPrescribingGuidance) {
+        this.otherPrescribingGuidance = otherPrescribingGuidance;
+        return this;
+    }
+
     public boolean isPediatric() {
         return pediatric;
     }