Parse protein change case insensitive

src/main/java/org/mskcc/oncokb/curation/service/MainService.java

@@ -149,7 +149,7 @@ public AlterationAnnotationStatus annotateAlteration(ReferenceGenome referenceGe
 
         // update associated genes
         Set<Gene> genes = alteration.getGenes();
-        if (parsedAlteration.getType().equals(STRUCTURAL_VARIANT) && !parsedAlteration.getGenes().isEmpty()) {
+        if (STRUCTURAL_VARIANT.equals(parsedAlteration.getType()) && !parsedAlteration.getGenes().isEmpty()) {
             genes = parsedAlteration.getGenes();
         }
         Set<Gene> annotatedGenes = genes
@@ -230,7 +230,8 @@ public AlterationAnnotationStatus annotateAlteration(ReferenceGenome referenceGe
                 ) {
                     String refRe = String.valueOf(canonicalSequenceOptional.orElseThrow().getSequence().charAt(alteration.getStart() - 1));
                     if (!StringUtils.isEmpty(refRe)) {
-                        if (StringUtils.isEmpty(alteration.getRefResidues())) {
+                        // only set the reference AA when the alteration happens on one position
+                        if (StringUtils.isEmpty(alteration.getRefResidues()) && alteration.getStart().equals(alteration.getEnd())) {
                             alteration.setRefResidues(refRe);
                         } else {
                             // If The AA in alteration is differed from the canonical transcript, and it's not X, we give warning

src/main/java/org/mskcc/oncokb/curation/util/parser/ParsingStatus.java

@@ -0,0 +1,38 @@
+package org.mskcc.oncokb.curation.util.parser;
+
+import org.mskcc.oncokb.curation.domain.enumeration.EntityStatusType;
+
+public class ParsingStatus<T> {
+
+    EntityStatusType status;
+    String message;
+    T entity;
+
+    public Boolean isParsed() {
+        return this.status != null;
+    }
+
+    public EntityStatusType getStatus() {
+        return status;
+    }
+
+    public void setStatus(EntityStatusType status) {
+        this.status = status;
+    }
+
+    public String getMessage() {
+        return message;
+    }
+
+    public void setMessage(String message) {
+        this.message = message;
+    }
+
+    public T getEntity() {
+        return entity;
+    }
+
+    public void setEntity(T entity) {
+        this.entity = entity;
+    }
+}

src/test/java/org/mskcc/oncokb/curation/TestHelper.java

@@ -0,0 +1,20 @@
+package org.mskcc.oncokb.curation;
+
+import java.io.BufferedReader;
+import java.io.File;
+import java.io.FileNotFoundException;
+import java.io.FileReader;
+
+public class TestHelper {
+
+    public static BufferedReader getTestFileBufferedReader(String filePath) throws FileNotFoundException {
+        if (filePath == null) {
+            System.out.println("Please specify the testing file path");
+            return null;
+        }
+
+        File file = new File(filePath);
+        FileReader reader = new FileReader(file);
+        return new BufferedReader(reader);
+    }
+}

src/test/java/org/mskcc/oncokb/curation/test/AlterationUtilsTest.java

@@ -0,0 +1,25 @@
+package org.mskcc.oncokb.curation.test;
+
+import static org.junit.Assert.assertEquals;
+import static org.mskcc.oncokb.curation.util.AlterationUtils.parseProteinChange;
+
+import org.junit.jupiter.api.Test;
+import org.mskcc.oncokb.curation.domain.Alteration;
+import org.mskcc.oncokb.curation.domain.AlterationAnnotationStatus;
+import org.mskcc.oncokb.curation.domain.EntityStatus;
+
+public class AlterationUtilsTest {
+
+    @Test
+    public void testRevisedProteinChangeInParseProteinChange() {
+        EntityStatus<Alteration> status = new AlterationAnnotationStatus();
+        parseProteinChange(status, "v600e");
+        assertEquals("V600E", status.getEntity().getProteinChange());
+
+        parseProteinChange(status, "*757kext*");
+        assertEquals("*757Kext*", status.getEntity().getProteinChange());
+
+        parseProteinChange(status, "T599delinsip");
+        assertEquals("T599delinsIP", status.getEntity().getProteinChange());
+    }
+}

src/test/java/org/mskcc/oncokb/curation/test/ParseProteinChangeParameterizedTest.java

@@ -0,0 +1,144 @@
+package org.mskcc.oncokb.curation.test;
+
+import static org.junit.Assert.assertEquals;
+import static org.mskcc.oncokb.curation.TestHelper.getTestFileBufferedReader;
+import static org.mskcc.oncokb.curation.util.AlterationUtils.parseProteinChange;
+
+import java.io.BufferedReader;
+import java.io.IOException;
+import java.util.ArrayList;
+import java.util.Collection;
+import java.util.List;
+import org.apache.commons.lang3.StringUtils;
+import org.junit.Test;
+import org.junit.runner.RunWith;
+import org.junit.runners.Parameterized;
+import org.mskcc.oncokb.curation.domain.Alteration;
+import org.mskcc.oncokb.curation.domain.AlterationAnnotationStatus;
+import org.mskcc.oncokb.curation.domain.EntityStatus;
+
+@RunWith(Parameterized.class)
+public class ParseProteinChangeParameterizedTest {
+
+    private static String TEST_FILE_PATH = "src/test/resources/data/test_parse_protein_change.tsv";
+
+    private String proteinChange;
+    private String expectedConsequence;
+    private String expectedRefAllele;
+    private String expectedVarAllele;
+    private String expectedProteinStart;
+    private String expectedProteinEnd;
+
+    public ParseProteinChangeParameterizedTest(
+        String proteinChange,
+        String expectedConsequence,
+        String expectedRefAllele,
+        String expectedVarAllele,
+        String expectedProteinStart,
+        String expectedProteinEnd
+    ) {
+        this.proteinChange = proteinChange;
+        this.expectedConsequence = expectedConsequence;
+        this.expectedRefAllele = expectedRefAllele;
+        this.expectedVarAllele = expectedVarAllele;
+        this.expectedProteinStart = expectedProteinStart;
+        this.expectedProteinEnd = expectedProteinEnd;
+    }
+
+    @Parameterized.Parameters
+    public static Collection<String[]> getParameters() throws IOException {
+        return importer();
+    }
+
+    private static List<String[]> importer() throws IOException {
+        BufferedReader buf = getTestFileBufferedReader(TEST_FILE_PATH);
+        String line = buf.readLine();
+
+        List<String[]> queries = new ArrayList<>();
+        int count = 0;
+        while (line != null) {
+            if (!line.startsWith("#") && line.trim().length() > 0) {
+                try {
+                    String parts[] = line.split("\t");
+                    if (parts.length < 1) {
+                        throw new IllegalArgumentException("Test case should have at least protein change. Current case: " + line);
+                    }
+                    String proteinChange = parts[0];
+                    String expectedConsequence = parts.length > 1 ? parts[1].toUpperCase() : "";
+                    String expectedRefAllele = parts.length > 2 ? parts[2] : "";
+                    String expectedVarAllele = parts.length > 3 ? parts[3] : "";
+                    String expectedProteinStart = parts.length > 4 ? parts[4] : "";
+                    String expectedProteinEnd = parts.length > 5 ? parts[5] : "";
+                    String[] query = {
+                        proteinChange,
+                        expectedConsequence,
+                        expectedRefAllele,
+                        expectedVarAllele,
+                        expectedProteinStart,
+                        expectedProteinEnd,
+                    };
+                    queries.add(query);
+                    count++;
+                } catch (Exception e) {
+                    System.err.println("Could not add line '" + line + "'. " + e);
+                }
+            }
+            line = buf.readLine();
+        }
+        System.err.println("Contains " + count + " queries.");
+        System.err.println("Done.");
+
+        return queries;
+    }
+
+    private void testSuite(
+        Alteration annotatedAlteration,
+        String proteinChange,
+        String expectedConsequence,
+        String expectedRefAllele,
+        String expectedVarAllele,
+        String expectedProteinStart,
+        String expectedProteinEnd
+    ) {
+        assertEquals(
+            "Not expected consequence. Query: " + proteinChange,
+            expectedConsequence,
+            annotatedAlteration.getConsequence() == null ? "" : annotatedAlteration.getConsequence().getTerm()
+        );
+        assertEquals(
+            "Not expected ref allele. Query: " + proteinChange,
+            expectedRefAllele,
+            StringUtils.isEmpty(annotatedAlteration.getRefResidues()) ? "" : annotatedAlteration.getRefResidues()
+        );
+        assertEquals(
+            "Not expected var allele. Query: " + proteinChange,
+            expectedVarAllele,
+            StringUtils.isEmpty(annotatedAlteration.getVariantResidues()) ? "" : annotatedAlteration.getVariantResidues()
+        );
+        assertEquals(
+            "Not expected protein start. Query: " + proteinChange,
+            expectedProteinStart,
+            annotatedAlteration.getStart() == null ? "" : Integer.toString(annotatedAlteration.getStart())
+        );
+        assertEquals(
+            "Not expected protein end. Query: " + proteinChange,
+            expectedProteinEnd,
+            annotatedAlteration.getEnd() == null ? "" : Integer.toString(annotatedAlteration.getEnd())
+        );
+    }
+
+    @Test
+    public void testSummary() {
+        EntityStatus<Alteration> alterationEntityStatus = new AlterationAnnotationStatus();
+        parseProteinChange(alterationEntityStatus, proteinChange);
+        testSuite(
+            alterationEntityStatus.getEntity(),
+            proteinChange,
+            expectedConsequence,
+            expectedRefAllele,
+            expectedVarAllele,
+            expectedProteinStart,
+            expectedProteinEnd
+        );
+    }
+}

src/test/resources/data/test_parse_protein_change.tsv

@@ -0,0 +1,92 @@
+#
+# List of alterations for testing.  Each line is the following format:
+#
+# Protein Change\tExpected consequence\tExpected Reference Allele\tExpected Variant Allele\tExpected Protein Start\tExpected Protein End
+#
+
+# missense variants
+N505I	missense_variant	N	I	505	505
+E323_D324delinsKN	missense_variant			323	324
+814_852mis	missense_variant			814	852
+IK744KI	missense_variant	IK	KI	744	745
+
+# inframes
+Q58_Q59insL	inframe_insertion			58	59
+58_Q59insL	inframe_insertion			58	59
+Q58_59insL	inframe_insertion			58	59
+58_59insL	inframe_insertion			58	59
+P68_C77dup	inframe_insertion			68	77
+P68_77dup	inframe_insertion			68	77
+68_C77dup	inframe_insertion			68	77
+68_77dup	inframe_insertion			68	77
+T599delinsIP	inframe_insertion	T		599	599
+599delinsIP	inframe_insertion			599	599
+599delins	UNKNOWN			599	599
+599delins2	UNKNOWN			599	599
+I744KI	inframe_insertion	I	KI	744	744
+M1ext-1	inframe_insertion			1	1
+1ext-1	inframe_insertion			1	1
+1ext	inframe_insertion			1	1
+C359del	inframe_deletion	C		359	359
+359del	inframe_deletion			359	359
+IK744A	inframe_deletion	IK	A	744	745
+1459_1468del	inframe_deletion			1459	1468
+A1459_D1468del	inframe_deletion			1459	1468
+1459_D1468del	inframe_deletion			1459	1468
+A1459_1468del	inframe_deletion			1459	1468
+D1161_S1172delinsE	inframe_deletion			1161	1172
+D1161_1172delinsE	inframe_deletion			1161	1172
+1161_S1172delinsE	inframe_deletion			1161	1172
+1161_1172delinsE	inframe_deletion			1161	1172
+
+# frame shifts
+S859Afs*12	frameshift_variant	S		859	859
+859Afs*12	frameshift_variant			859	859
+S859fs*12	frameshift_variant	S		859	859
+G314fs	frameshift_variant	G		314	314
+314fs	frameshift_variant			314	314
+S330_S352fs	frameshift_variant			330	352
+330_S352fs	frameshift_variant			330	352
+S330_352fs	frameshift_variant			330	352
+330_352fs	frameshift_variant			330	352
+
+#  truncating mutations
+W143_A314trunc	feature_truncation			143	314
+W143_314trunc	feature_truncation			143	314
+143_A314trunc	feature_truncation			143	314
+422_605trunc	feature_truncation			422	605
+Truncating Mutations	feature_truncation
+
+# splice variants
+596_619splice	splice_region_variant			596	619
+X1429_splice	splice_region_variant			1429	1429
+1429_splice	splice_region_variant			1429	1429
+M1I	start_lost	M	I	1	1
+1I	start_lost		I	1	1
+M1?	start_lost	M	?	1	1
+A149*	stop_gained	A	*	149	149
+149*	stop_gained		*	149	149
+*149*	stop_lost	*	*	149	149
+
+# extension
+*757Kext*36	stop_lost	*		757	757
+757Kext*36	stop_lost	*		757	757
+*757ext*36	stop_lost	*		757	757
+*757Kext*	stop_lost	*		757	757
+*757Kext*?	stop_lost	*		757	757
+
+# synonymous variants
+K24K	synonymous_variant	K	K	24	24
+
+# any
+449_514mut	any			449	514
+
+# Unknown variants
+D399	NA	D		399	399
+MCUR1-AKT1 fusion	UNKNOWN
+Fusions	UNKNOWN
+Deletion	UNKNOWN
+Amplification	UNKNOWN
+Oncogenic Mutations	UNKNOWN
+Oncogenic Mutations {excluding V600}	UNKNOWN
+