fix(phenotyper): fix bug with stripping gene symbol from outside call…

… diplotypes Remove astrolabe support. Fix-for: #161
PharmGKB · Nov 15, 2023 · cb97898 · cb97898
1 parent bcc2c81
commit cb97898
Show file tree

Hide file tree

Showing 7 changed files with 73 additions and 46 deletions.
diff --git a/docs/using/Outside-Call-Format.md b/docs/using/Outside-Call-Format.md
@@ -73,6 +73,9 @@ We rely on string matching to match outside calls to recommendations.
 
 Consult the [Phenotypes List](/Phenotypes-List) for a complete list of named alleles, phenotypes and activity scores.
 
+_Do not_ prefix allele names with the gene symbol in the second field (e.g. CYP2C9*1/CYP2C9*3). The gene is specified in
+the first field so repeating it in second field is not necessary.
+
 If there is an outside call for a gene that also has data from the VCF, the outside call will trump the VCF data.
 
 

diff --git a/src/main/java/org/pharmgkb/pharmcat/phenotype/model/OutsideCall.java b/src/main/java/org/pharmgkb/pharmcat/phenotype/model/OutsideCall.java
@@ -13,7 +13,7 @@
 
 
 /**
- * Calls from an outside data source. For example, Astrolabe (but could be others)
+ * Calls from an outside data source.
  *
  * @author Ryan Whaley
  */
@@ -51,7 +51,7 @@ public OutsideCall(String line, int lineNumber) throws RuntimeException {
       throw new BadOutsideCallException("Line " + lineNumber + ": No gene specified");
     }
 
-    m_diplotype = StringUtils.stripToNull(fields.get(IDX_DIPS).replaceAll(m_gene, ""));
+    m_diplotype = StringUtils.stripToNull(fields.get(IDX_DIPS));
     if (fields.size() == 2 && (m_diplotype == null || m_diplotype.equals(sf_diplotypeSeparator))) {
       if (StringUtils.isBlank(fields.get(IDX_DIPS))) {
         throw new BadOutsideCallException("Line " + lineNumber + ": No diplotype specified");

diff --git a/src/test/java/org/pharmgkb/pharmcat/PipelineTest.java b/src/test/java/org/pharmgkb/pharmcat/PipelineTest.java
@@ -67,7 +67,7 @@ static void prepare() throws IOException {
     try (BufferedWriter writer = Files.newBufferedWriter(s_outsideCallFilePath)) {
       writer.write("""
           ##Test Outside Call Data
-          CYP2D6\tCYP2D6*1/CYP2D6*4\t\t\t0.6\t0.75\tp: 0.0\t\t\tv1.9-2017_02_09
+          CYP2D6\t*1/*4\t\t\t0.6\t0.75\tp: 0.0\t\t\tv1.9-2017_02_09
               """);
     }
 
@@ -797,6 +797,31 @@ void testCftrD1270NG551D(TestInfo testInfo) throws Exception {
   }
 
 
+  /**
+   *  Check to see if the CPIC-specified allele name of "ivacaftor non-responsive CFTR sequence" gets translated to the
+   *  more generic value of "Reference" for display in the report
+   */
+  @Test
+  void testCftrOutsideReferenceCall(TestInfo testInfo) throws Exception {
+    Path outsideCallPath = TestUtils.createTestFile(testInfo, ".tsv");
+    try (PrintWriter writer = new PrintWriter(Files.newBufferedWriter(outsideCallPath))) {
+      writer.println("CFTR\tivacaftor non-responsive CFTR sequence/ivacaftor non-responsive CFTR sequence");
+    }
+    PipelineWrapper testWrapper = new PipelineWrapper(testInfo, false);
+    testWrapper.getVcfBuilder()
+        .reference("CYP2C19")
+        .reference("CYP2C9")
+    ;
+    testWrapper.execute(outsideCallPath);
+
+    testWrapper.testCalledByMatcher("CYP2C19");
+    testWrapper.testCalledByMatcher("CYP2C9");
+
+    testWrapper.testReportable("CFTR");
+    testWrapper.testPrintCalls(DataSource.CPIC, "CFTR", "Reference/Reference");
+  }
+
+
   @Test
   void testRosuvastatin(TestInfo testInfo) throws Exception {
     PipelineWrapper testWrapper = new PipelineWrapper(testInfo, false);

diff --git a/src/test/java/org/pharmgkb/pharmcat/phenotype/OutsideCallParserTest.java b/src/test/java/org/pharmgkb/pharmcat/phenotype/OutsideCallParserTest.java
@@ -15,35 +15,14 @@
 
 class OutsideCallParserTest {
 
-  @Test
-  void testAstrolabeFormat(TestInfo testInfo) throws IOException {
-    String sf_astrolabeOutput = "##Test Astrolabe output\n" +
-        "#ROI_label\tdiplotype labels\tdiplotype activity\tdiplotype calling notes\tjaccard\tpart\tpValue\tROI notes\t" +
-        "special case\tnomenclature version\n" +
-        "CYP2D6\tCYP2D6*1/CYP2D6*4\t\t\t0.6\t0.75\tp: 0.0\t\t\tv1.9-2017_02_09\n";
-
-    Path tempAstroPath = TestUtils.createTestFile(testInfo, ".tsv");
-    try (FileWriter fw = new FileWriter(tempAstroPath.toFile())) {
-      fw.write(sf_astrolabeOutput);
-    }
-
-    List<OutsideCall> calls = OutsideCallParser.parse(tempAstroPath);
-    assertNotNull(calls);
-    assertEquals(1, calls.size());
-    assertEquals("CYP2D6", calls.get(0).getGene());
-    assertEquals("*1/*4", calls.get(0).getDiplotype());
-  }
-
   @Test
   void testMinimalInput(TestInfo testInfo) throws IOException {
-    String sf_astrolabeOutput = "CYP2C9\t*1/*2";
-
-    Path tempAstroPath = TestUtils.createTestFile(testInfo, ".tsv");
-    try (FileWriter fw = new FileWriter(tempAstroPath.toFile())) {
-      fw.write(sf_astrolabeOutput);
+    Path outsideCallPath = TestUtils.createTestFile(testInfo, ".tsv");
+    try (FileWriter fw = new FileWriter(outsideCallPath.toFile())) {
+      fw.write("CYP2C9\t*1/*2");
     }
 
-    List<OutsideCall> calls = OutsideCallParser.parse(tempAstroPath);
+    List<OutsideCall> calls = OutsideCallParser.parse(outsideCallPath);
     assertNotNull(calls);
     assertEquals(1, calls.size());
     assertEquals("CYP2C9", calls.get(0).getGene());
@@ -52,14 +31,15 @@ void testMinimalInput(TestInfo testInfo) throws IOException {
 
   @Test
   void testTwoGenes(TestInfo testInfo) throws IOException {
-    String sf_astrolabeOutput = "CYP2C9\t*1/*2\nCYP2C19\t*3/*4";
-
-    Path tempAstroPath = TestUtils.createTestFile(testInfo, ".tsv");
-    try (FileWriter fw = new FileWriter(tempAstroPath.toFile())) {
-      fw.write(sf_astrolabeOutput);
+    Path tmpOutsideCallPath = TestUtils.createTestFile(testInfo, ".tsv");
+    try (FileWriter fw = new FileWriter(tmpOutsideCallPath.toFile())) {
+      fw.write("""
+          CYP2C9\t*1/*2
+          CYP2C19\t*3/*4
+          """);
     }
 
-    List<OutsideCall> calls = OutsideCallParser.parse(tempAstroPath);
+    List<OutsideCall> calls = OutsideCallParser.parse(tmpOutsideCallPath);
     assertNotNull(calls);
     assertEquals(2, calls.size());
     assertEquals("CYP2C9", calls.get(0).getGene());
@@ -71,15 +51,38 @@ void testTwoGenes(TestInfo testInfo) throws IOException {
 
   @Test
   void testBadFormat(TestInfo testInfo) throws IOException {
-    String sf_astrolabeOutput = "CYP2C9\t*1/*2\nCYP2C19\t*3/*4/*2";
-
-    Path tempAstroPath = TestUtils.createTestFile(testInfo, ".tsv");
-    try (FileWriter fw = new FileWriter(tempAstroPath.toFile())) {
-      fw.write(sf_astrolabeOutput);
+    Path outsideCallPath = TestUtils.createTestFile(testInfo, ".tsv");
+    try (FileWriter fw = new FileWriter(outsideCallPath.toFile())) {
+      fw.write("""
+          CYP2C9\t*1/*2
+          CYP2C19\t*3/*4/*2
+          """);
     }
 
     assertThrows(BadOutsideCallException.class, () -> {
-      OutsideCallParser.parse(tempAstroPath);
+      OutsideCallParser.parse(outsideCallPath);
     });
   }
+
+
+  @Test
+  void testCommentsAndEmptyLines(TestInfo testInfo) throws IOException {
+    Path outsideCallPath = TestUtils.createTestFile(testInfo, ".tsv");
+    try (FileWriter fw = new FileWriter(outsideCallPath.toFile())) {
+      fw.write("""
+          # comment
+          CYP2C9\t*1/*2
+          
+          ## another comment
+          
+          CYP2C9\t*3/*4
+          
+          """);
+    }
+
+    List<OutsideCall> calls = OutsideCallParser.parse(outsideCallPath);
+    assertEquals(2, calls.size());
+    assertEquals("*1/*2", calls.get(0).getDiplotype());
+    assertEquals("*3/*4", calls.get(1).getDiplotype());
+  }
 }
diff --git a/src/test/java/org/pharmgkb/pharmcat/phenotype/model/OutsideCallTest.java b/src/test/java/org/pharmgkb/pharmcat/phenotype/model/OutsideCallTest.java
@@ -36,7 +36,7 @@ void testBad() {
 
     assertThrows(BadOutsideCallException.class, () -> {
       try {
-        new OutsideCall("CYP2D6\tCYP2D6", 3);
+        new OutsideCall("CYP2D6\t   ", 3);
       } catch (BadOutsideCallException ex) {
         System.out.println(ex.getMessage());
         throw ex;

diff --git a/src/test/resources/org/pharmgkb/pharmcat/reporter/outside_CYP2D6.tsv b/src/test/resources/org/pharmgkb/pharmcat/reporter/outside_CYP2D6.tsv
@@ -1,3 +1 @@
-##Test Astrolabe output
-#ROI_label	diplotype labels	diplotype activity	diplotype calling notes	jaccard	part	pValue	ROI notes	special case	nomenclature version
 CYP2D6	CYP2D6*3/CYP2D6*4
diff --git a/src/test/resources/org/pharmgkb/pharmcat/reporter/outside_CYP2D6_G6PD.tsv b/src/test/resources/org/pharmgkb/pharmcat/reporter/outside_CYP2D6_G6PD.tsv
@@ -1,4 +1,2 @@
-##Test Astrolabe output
-#ROI_label	diplotype labels	diplotype activity	diplotype calling notes	jaccard	part	pValue	ROI notes	special case	nomenclature version
 CYP2D6	CYP2D6*3/CYP2D6*4
-G6PD	B (wildtype)/B (wildtype)
+G6PD	B (wildtype)/B (wildtype)