fix: validate outside call diplotypes

PharmGKB · Aug 14, 2024 · 9c49fbc · 9c49fbc
1 parent 0cbb7d0
commit 9c49fbc
Show file tree

Hide file tree

Showing 4 changed files with 77 additions and 11 deletions.
diff --git a/src/main/java/org/pharmgkb/pharmcat/Env.java b/src/main/java/org/pharmgkb/pharmcat/Env.java
@@ -4,8 +4,12 @@
 import java.nio.file.Path;
 import java.util.HashMap;
 import java.util.Map;
+import java.util.Optional;
+import com.google.common.collect.HashMultimap;
+import com.google.common.collect.Multimap;
 import org.checkerframework.checker.nullness.qual.Nullable;
 import org.pharmgkb.pharmcat.definition.DefinitionReader;
+import org.pharmgkb.pharmcat.definition.model.DefinitionFile;
 import org.pharmgkb.pharmcat.phenotype.PhenotypeMap;
 import org.pharmgkb.pharmcat.phenotype.model.GenePhenotype;
 import org.pharmgkb.pharmcat.reporter.MessageHelper;
@@ -28,6 +32,7 @@ public class Env {
   private final PgkbGuidelineCollection m_drugs;
   private MessageHelper m_messageHelper;
   private final Map<DataSource, Map<String, Map<String, Haplotype>>> m_haplotypeCache = new HashMap<>();
+  private final Multimap<String, String> m_validHaplotypes = HashMultimap.create();
 
 
   public Env() throws IOException, ReportableException {
@@ -74,6 +79,50 @@ public String getReferenceAllele(String gene) {
   }
 
 
+  /**
+   * Checks if the specified allele is used in either definition files or phenotype.
+   */
+  public boolean isValidNamedAllele(String gene, String allele) {
+
+    if (m_validHaplotypes.containsEntry(gene, allele)) {
+      return true;
+    }
+
+    if (gene.startsWith("HLA-")) {
+      // HLA's are a special case
+      m_validHaplotypes.put(gene, allele);
+      return true;
+    }
+
+    Optional<DefinitionFile> opt = m_definitionReader.lookupDefinitionFile(gene);
+    if (opt.isPresent() && opt.get().getNamedAllele(allele) != null) {
+      m_validHaplotypes.put(gene, allele);
+      return true;
+    }
+
+    String inferredAllele = allele;
+    if (gene.equals("CYP2D6")) {
+      inferredAllele = Cyp2d6CopyNumberCaller.inferHaplotypeName(allele);
+    }
+    GenePhenotype gp = m_phenotypeMap.getPhenotype(gene, DataSource.CPIC);
+    if (gp != null) {
+      if (gp.getHaplotypes().containsKey(inferredAllele) || gp.getActivityValues().containsKey(inferredAllele)) {
+        m_validHaplotypes.put(gene, allele);
+        return true;
+      }
+    }
+    gp = m_phenotypeMap.getPhenotype(gene, DataSource.DPWG);
+    if (gp != null) {
+      boolean rez = gp.getHaplotypes().containsKey(inferredAllele) || gp.getActivityValues().containsKey(inferredAllele);
+      if (rez) {
+        m_validHaplotypes.put(gene, allele);
+      }
+      return rez;
+    }
+    return false;
+  }
+
+
   public PgkbGuidelineCollection getDrugs() {
     return m_drugs;
   }

diff --git a/src/main/java/org/pharmgkb/pharmcat/Pipeline.java b/src/main/java/org/pharmgkb/pharmcat/Pipeline.java
@@ -300,6 +300,11 @@ public PipelineResult call() throws IOException {
                   output.add(AnsiConsole.styleWarning(msg));
                 }
               }
+              for (String hap : call.getHaplotypes()) {
+                if (!m_env.isValidNamedAllele(gene, hap)) {
+                  call.addWarning("Undocumented " + gene + " named allele in outside call: " + hap);
+                }
+              }
               outsideCalls.add(call);
               call.getWarnings().forEach(w -> output.add(AnsiConsole.styleWarning("WARNING: " + w)));
               outsideCallMap.computeIfAbsent(gene, g -> new HashSet<>()).add(call);

diff --git a/src/main/java/org/pharmgkb/pharmcat/phenotype/model/OutsideCall.java b/src/main/java/org/pharmgkb/pharmcat/phenotype/model/OutsideCall.java
@@ -77,6 +77,9 @@ public OutsideCall(String line, int lineNumber) throws RuntimeException {
       List<String> alleles = sf_diplotypeSplitter.splitToList(diplotype).stream()
           .map(a -> a.replaceFirst("^" + m_gene + "\\s*", ""))
           .toList();
+      if (alleles.size() > 2) {
+        throw new BadOutsideCallException("Line " + lineNumber + ": Too many alleles specified in " + diplotype);
+      }
 
       if (m_gene.equals("CYP2D6")) {
         alleles = alleles.stream()
@@ -121,12 +124,10 @@ public OutsideCall(String line, int lineNumber) throws RuntimeException {
             .toList();
       }
 
-      // re-join alleles to eliminate white space when gene symbol is used in diplotype
+      // re-join alleles to eliminate white space when a gene symbol is used in diplotype
       m_diplotype = String.join(sf_diplotypeSeparator, alleles);
       m_diplotypes = ImmutableList.of(m_diplotype);
-      if (alleles.size() > 2) {
-        throw new BadOutsideCallException("Line " + lineNumber + ": Too many alleles specified in " + m_diplotype);
-      }
+
       m_haplotypes.add(alleles.get(0));
       if (alleles.size() == 2) {
         m_haplotypes.add(alleles.get(1));
@@ -146,6 +147,10 @@ public OutsideCall(String line, int lineNumber) throws RuntimeException {
     }
   }
 
+  public void addWarning(String warning) {
+    m_warnings.add(warning);
+  }
+
   @Override
   public String toString() {
     if (m_diplotype != null) {

diff --git a/src/main/java/org/pharmgkb/pharmcat/reporter/caller/Cyp2d6CopyNumberCaller.java b/src/main/java/org/pharmgkb/pharmcat/reporter/caller/Cyp2d6CopyNumberCaller.java
@@ -87,24 +87,31 @@ private static Object[] inferHaplotype(Haplotype haplotype) {
     if (haplotype == null) {
       return new Object[] {false, null};
     }
-    Matcher m = sf_copyNumberPattern.matcher(haplotype.getName());
+    String name = inferHaplotypeName(haplotype.getName());
+    return new Object[] {!name.equals(haplotype.getName()), name};
+  }
+
+
+  public static String inferHaplotypeName(String haplotypeName) {
+
+    Matcher m = sf_copyNumberPattern.matcher(haplotypeName);
     if (!m.matches()) {
-      return new Object[] {false, haplotype.getName()};
+      return haplotypeName;
     }
     Integer hap = Integer.parseInt(m.group(1));
     if (!m_gteThree.contains(hap)) {
-      return new Object[] {false, haplotype.getName()};
+      return haplotypeName;
     }
     int cn = Integer.parseInt(m.group(3));
     if (cn <= 2) {
-      return new Object[] {false, haplotype.getName()};
+      return haplotypeName;
     }
-    if (haplotype.getName().contains(TextConstants.GTE)) {
+    if (haplotypeName.contains(TextConstants.GTE)) {
       if (cn == 3) {
         // ignore >= 3
-        return new Object[] {false, haplotype.getName()};
+        return haplotypeName;
       }
     }
-    return new Object[] {true, "*" + hap + "x" + TextConstants.GTE + "3"};
+    return "*" + hap + "x" + TextConstants.GTE + "3";
   }
 }