Merge db91f04 into 701301b

.github/workflows/ci.yml

@@ -81,6 +81,10 @@ jobs:
 
     steps:
       - uses: actions/checkout@v2
+        with:
+          lfs: true
+      - name: Check out LFS objects
+        run: git lfs fetch --all
       - name: Set up JDK
         uses: actions/setup-java@v2
         with:

tests/hg19-chr22/Case_1_index.delly2.feature-effects.tsv-expected

@@ -1 +1,3 @@
 case_id	set_id	sv_uuid	refseq_gene_id	refseq_transcript_id	refseq_transcript_coding	refseq_effect	ensembl_gene_id	ensembl_transcript_id	ensembl_transcript_coding	ensembl_effect
+.	.	UUID	51816	NM_017424.2	TRUE	{"transcript_ablation","coding_transcript_variant"}	ENSG00000093072	ENST00000262607.3	TRUE	{"transcript_ablation","coding_transcript_variant"}
+.	.	UUID	128954	NM_001037814.1	TRUE	{"transcript_ablation","coding_transcript_variant"}	ENSG00000215568	ENST00000400588.1	TRUE	{"transcript_ablation","coding_transcript_variant"}

tests/hg19-chr22/Case_1_index.delly2.gts.tsv-expected

@@ -1 +1,2 @@
 release	chromosome	chromosome_no	bin	chromosome2	chromosome_no2	bin2	pe_orientation	start	end	start_ci_left	start_ci_right	end_ci_left	end_ci_right	case_id	set_id	sv_uuid	caller	sv_type	sv_sub_type	info	num_hom_alt	num_hom_ref	num_het	num_hemi_alt	num_hemi_ref	genotype
+GRCh37	22	22	89	22	22	89	3to5	17400000	17700000	-29	29	-29	29	.	.	UUID	EMBL.DELLYv1.1.3	DEL	DEL	{"""backgroundCarriers""":0,"""affectedCarriers""":0,"""unaffectedCarriers""":0}	0	2	1	0	0	{"""Case_1_father-N1-DNA1-WGS1""":{"""gt""":"""0/1""","""gq""":14,"""pec""":0,"""pev""":0,"""src""":34,"""srv""":4},"""Case_1_index-N1-DNA1-WGS1""":{"""gt""":"""0/1""","""gq""":14,"""pec""":0,"""pev""":0,"""src""":34,"""srv""":4,"""gt""":"""0/0""","""gq""":35,"""pec""":0,"""pev""":0,"""src""":29,"""srv""":2},"""Case_1_mother-N1-DNA1-WGS1""":{"""gt""":"""0/1""","""gq""":14,"""pec""":0,"""pev""":0,"""src""":34,"""srv""":4,"""gt""":"""0/0""","""gq""":35,"""pec""":0,"""pev""":0,"""src""":29,"""srv""":2,"""gt""":"""0/0""","""gq""":67,"""pec""":0,"""pev""":0,"""src""":32,"""srv""":1}}

tests/hg19-chr22/build.sh

@@ -110,9 +110,12 @@ VCF=$BASEDIR/wgs_sv_calling/output/bwa.delly2.Case_1_index-N1-DNA1-WGS1/out/bwa.
 
 ( \
   tabix --only-header $VCF $REGIONS; \
-  tabix $VCF $REGIONS \
+  ( \
+    cat Case_1_index.delly2.extra.vcf; \
+    tabix $VCF $REGIONS; \
+  ) \
   | sort -k1,1V -k2,2n \
-  | uniq; \
+  | uniq; \b
 ) \
 | bgzip -c \
 > Case_1_index.delly2.vcf.gz

tests/hg19-chr22/run-tests.sh

@@ -7,7 +7,7 @@ JAR=$(ls ../../varfish-annotator-cli/target/varfish-annotator-cli-*.jar | grep -
 
 ## prepare
 
-gzip -d -k chr22_part.fa.gz
+gzip -d chr22_part.fa.gz > chr22_part.fa
 
 ## step 0: help
 
@@ -84,9 +84,13 @@ java -jar $JAR annotate-svs \
   --db-path /tmp/out.h2.db \
   --self-test-chr22-only
 
-diff /tmp/Case_1_index.delly2.gts.tsv Case_1_index.delly2.gts.tsv-expected
+awk -F $'\t' 'BEGIN { OFS=FS } { if (NR > 1) $17 = "UUID"; print $0; }' /tmp/Case_1_index.delly2.gts.tsv \
+> /tmp/Case_1_index.delly2.gts.tsv.replaced
+diff /tmp/Case_1_index.delly2.gts.tsv.replaced Case_1_index.delly2.gts.tsv-expected
 diff /tmp/Case_1_index.delly2.db-info.tsv Case_1_index.delly2.db-info.tsv-expected
-diff /tmp/Case_1_index.delly2.feature-effects.tsv Case_1_index.delly2.feature-effects.tsv-expected
+awk -F $'\t' 'BEGIN { OFS=FS } { if (NR > 1) $3 = "UUID"; print $0; }' /tmp/Case_1_index.delly2.feature-effects.tsv \
+> /tmp/Case_1_index.delly2.feature-effects.tsv.replaced
+diff /tmp/Case_1_index.delly2.feature-effects.tsv.replaced Case_1_index.delly2.feature-effects.tsv-expected
 
 ## if we reach here, everything is fine
 

varfish-annotator-cli/src/main/java/com/github/bihealth/varfish_annotator/annotate/IncompatibleVcfException.java

@@ -3,15 +3,7 @@
 /** Raised on incompatible VCF files. */
 public class IncompatibleVcfException extends Exception {
 
-  public IncompatibleVcfException(String message, Throwable cause) {
-    super(message, cause);
-  }
-
   public IncompatibleVcfException(String message) {
     super(message);
   }
-
-  public IncompatibleVcfException(Throwable cause) {
-    super(cause);
-  }
 }

varfish-annotator-cli/src/main/java/com/github/bihealth/varfish_annotator/annotate_svs/AnnotateSvsArgs.java

@@ -168,34 +168,18 @@ public String getDefaultSvMethod() {
     return defaultSvMethod;
   }
 
-  public void setDefaultSvMethod(String defaultSvMethod) {
-    this.defaultSvMethod = defaultSvMethod;
-  }
-
   public Boolean getSequentialUuids() {
     return sequentialUuids;
   }
 
-  public void setSequentialUuids(Boolean sequentialUuids) {
-    this.sequentialUuids = sequentialUuids;
-  }
-
   public String getOptOutFeatures() {
     return optOutFeatures;
   }
 
-  public void setOptOutFeatures(String optOutFeatures) {
-    this.optOutFeatures = optOutFeatures;
-  }
-
   public String getWriteBndMates() {
     return writeBndMates;
   }
 
-  public void setWriteBndMates(String writeBndMates) {
-    this.writeBndMates = writeBndMates;
-  }
-
   @Override
   public String toString() {
     return "AnnotateSvsArgs{"

varfish-annotator-cli/src/main/java/com/github/bihealth/varfish_annotator/annotate_svs/AnnotateSvsVcf.java

@@ -696,7 +696,7 @@ private String buildInfoValue(
       VariantContext ctx, GenomeVersion genomeVersion, SVGenomeVariant svGenomeVar) {
     final List<String> mappings = new ArrayList<>();
 
-    if (args.getOptOutFeatures().contains(FEATURE_SUPPRESS_CARRIER_COUNTS)) {
+    if (!args.getOptOutFeatures().contains(FEATURE_SUPPRESS_CARRIER_COUNTS)) {
       if (svGenomeVar.getType() == Type.BND) {
         final String contigName2 =
             (genomeVersion == GenomeVersion.HG19)

varfish-annotator-cli/src/main/java/com/github/bihealth/varfish_annotator/dbstats/DbStatsArgs.java

@@ -27,26 +27,14 @@ public boolean isHelp() {
     return help;
   }
 
-  public void setHelp(boolean help) {
-    this.help = help;
-  }
-
   public String getDbPath() {
     return dbPath;
   }
 
-  public void setDbPath(String dbPath) {
-    this.dbPath = dbPath;
-  }
-
   public boolean isParseable() {
     return parseable;
   }
 
-  public void setParseable(boolean parseable) {
-    this.parseable = parseable;
-  }
-
   @Override
   public String toString() {
     return "DbStatsArgs{"

varfish-annotator-cli/src/main/java/com/github/bihealth/varfish_annotator/utils/SelfTestFailedException.java

@@ -10,8 +10,4 @@ public SelfTestFailedException(String message, Throwable cause) {
   public SelfTestFailedException(String message) {
     super(message);
   }
-
-  public SelfTestFailedException(Throwable cause) {
-    super(cause);
-  }
 }

varfish-annotator-cli/src/main/java/com/github/bihealth/varfish_annotator/utils/UcscBinning.java

@@ -73,60 +73,6 @@ public static List<Integer> getOverlappingBins(int begin, int end) {
     return result;
   }
 
-  /**
-   * Given an (0-based, half-open) interval {@code begin:end} return all bins that completely
-   * contain {@code begin:end}.
-   *
-   * @param begin interval start position
-   * @param end interval end position
-   * @return the smallest bin that contains {@code begin:end}
-   */
-  public static List<Integer> getContainingBins(int begin, int end) {
-    final ArrayList<Integer> result = new ArrayList<>();
-    final int maxBin = getContainingBin(begin, end);
-    for (int bin : getOverlappingBins(begin, end)) {
-      if (bin <= maxBin) {
-        result.add(bin);
-      }
-    }
-    return result;
-  }
-
-  /**
-   * Given an (0-based, half-open) interval {@code begin:end} return all bins that are completely
-   * contained by {@code begin:end}.
-   *
-   * @param begin interval start position
-   * @param end interval end position
-   * @return the smallest bin that contains {@code begin:end}
-   */
-  public static List<Integer> getContainedBins(int begin, int end) {
-    final ArrayList<Integer> result = new ArrayList<>();
-    final int minBin = getContainingBin(begin, end);
-    for (int bin : getOverlappingBins(begin, end)) {
-      if (bin >= minBin) {
-        result.add(bin);
-      }
-    }
-    return result;
-  }
-
-  /** Return (0-based, half-open) interval that {@code bin} covers. */
-  public static int[] getCoveredBin(int bin) {
-    if (bin < 0 || bin > maxBin) {
-      throw new RuntimeException("Invalid bin number " + bin + ", max bin is " + maxBin);
-    }
-    int shift = binFirstShift;
-    for (int offset : binOffsets) {
-      if (offset <= bin) {
-        final int[] result = {(bin - offset) << shift, (bin + 1 - offset) << shift};
-        return result;
-      }
-      shift += binNextShift;
-    }
-    throw new RuntimeException("Unexpected error when computing covered bins");
-  }
-
   /**
    * Helper class for generating the first and last bins overlapping an interval {@code begin:end}.
    */

varfish-annotator-cli/src/test/java/com/github/bihealth/varfish_annotator/VarFishAnnotatorExceptionTest.java

@@ -0,0 +1,26 @@
+package com.github.bihealth.varfish_annotator;
+
+import org.junit.jupiter.api.Assertions;
+import org.junit.jupiter.api.Test;
+
+/** @author <a href="mailto:manuel.holtgrewe@bih-charite.de">Manuel Holtgrewe</a> */
+public class VarFishAnnotatorExceptionTest {
+
+  @Test
+  public void testConstructorString() {
+    Assertions.assertThrows(
+        VarfishAnnotatorException.class,
+        () -> {
+          throw new VarfishAnnotatorException("Test Message");
+        });
+  }
+
+  @Test
+  public void testConstructorStringThrowable() {
+    Assertions.assertThrows(
+        VarfishAnnotatorException.class,
+        () -> {
+          throw new VarfishAnnotatorException("Test Message", new Exception());
+        });
+  }
+}

varfish-annotator-cli/src/test/java/com/github/bihealth/varfish_annotator/annotate/AnnotateGatkHcVcf37Test.java → varfish-annotator-cli/src/test/java/com/github/bihealth/varfish_annotator/annotate/AnnotateGatkHcVcf37Chr1Test.java

@@ -20,7 +20,7 @@
 import org.junit.jupiter.params.provider.ValueSource;
 
 /** Test annotation of VCF files generated with GATK HC for GRCh37 */
-public class AnnotateGatkHcVcf37Test {
+public class AnnotateGatkHcVcf37Chr1Test {
 
   @TempDir public File tmpFolder;
   File h2DbFile;
@@ -37,11 +37,11 @@ void initEach() {
     ensemblSerFile = new File(tmpFolder + "/hg19_ensembl.ser");
     refseqSerFile = new File(tmpFolder + "/hg19_refseq_curated.ser");
 
-    ResourceUtils.gunzipResourceToFile("/grch37/hs37d5.1.fa.gz", fastaFile);
-    ResourceUtils.copyResourceToFile("/grch37/hs37d5.1.fa.fai", faiFile);
-    ResourceUtils.copyResourceToFile("/grch37/small-grch37.h2.db", h2DbFile);
-    ResourceUtils.copyResourceToFile("/grch37/hg19_ensembl.ser", ensemblSerFile);
-    ResourceUtils.copyResourceToFile("/grch37/hg19_refseq_curated.ser", refseqSerFile);
+    ResourceUtils.gunzipResourceToFile("/grch37-chr1/hs37d5.1.fa.gz", fastaFile);
+    ResourceUtils.copyResourceToFile("/grch37-chr1/hs37d5.1.fa.fai", faiFile);
+    ResourceUtils.copyResourceToFile("/grch37-chr1/small-grch37.h2.db", h2DbFile);
+    ResourceUtils.copyResourceToFile("/grch37-chr1/hg19_ensembl.ser", ensemblSerFile);
+    ResourceUtils.copyResourceToFile("/grch37-chr1/hg19_refseq_curated.ser", refseqSerFile);
   }
 
   void runTest(
@@ -136,7 +136,7 @@ void testWithSingleton(boolean gzipOutput) throws IOException {
     runTest(
         "bwa.gatk_hc.HG00102.vcf.gz",
         null,
-        "input/grch37",
+        "input/grch37-chr1",
         expectedDbInfo,
         expectedGts,
         gzipOutput,
@@ -170,7 +170,7 @@ void testWithTrio(boolean gzipOutput) throws IOException {
     runTest(
         "bwa.gatk_hc.NA12878.vcf.gz",
         null,
-        "input/grch37",
+        "input/grch37-chr1",
         expectedDbInfo,
         expectedGts,
         gzipOutput,
@@ -194,7 +194,7 @@ void testWithAsteriskAllele() throws IOException {
     runTest(
         "bwa.gatk_hc.HG00102.asterisk_alleles.vcf.gz",
         null,
-        "input/grch37",
+        "input/grch37-chr1",
         expectedDbInfo,
         expectedGts,
         false,
@@ -207,7 +207,14 @@ void testSelfTestFails() throws Exception {
     final String text =
         SystemLambda.tapSystemErr(
             () -> {
-              runTest("bwa.gatk_hc.NA12878.vcf.gz", null, "input/grch37", null, null, false, false);
+              runTest(
+                  "bwa.gatk_hc.NA12878.vcf.gz",
+                  null,
+                  "input/grch37-chr1",
+                  null,
+                  null,
+                  false,
+                  false);
             });
     Assertions.assertTrue(text.contains("Problem with database self-test:"));
   }
@@ -230,7 +237,7 @@ void testAnnotateHemiMale() throws Exception {
     runTest(
         "bwa.gatk_hc.HG00102.hemi.vcf.gz",
         "FAM_HG00102.ped",
-        "input/grch37",
+        "input/grch37-chr1",
         expectedDbInfo,
         expectedGts,
         false,
@@ -255,7 +262,7 @@ void testAnnotateHemiFamily() throws Exception {
     runTest(
         "bwa.gatk_hc.NA12878.hemi.vcf.gz",
         "FAM_NA12878.ped",
-        "input/grch37",
+        "input/grch37-chr1",
         expectedDbInfo,
         expectedGts,
         false,