rest of comment edits

broadinstitute · Mar 14, 2019 · 33f34b9 · 33f34b9
1 parent 5ca8d0b
commit 33f34b9
Show file tree

Hide file tree

Showing 4 changed files with 28 additions and 24 deletions.
diff --git a/scripts/mutect2_wdl/mutect2.wdl b/scripts/mutect2_wdl/mutect2.wdl
@@ -228,7 +228,7 @@ workflow Mutect2 {
         call LearnReadOrientationModel {
             input:
                 alt_tables = CollectF1R2Counts.alt_table,
-                ref_histograms = CollectF1R2Counts.ref_histogram,
+                ref_histograms = CollectF1R2Counts.ref_histograms,
                 alt_histograms = CollectF1R2Counts.alt_histograms,
                 gatk_override = gatk_override,
                 gatk_docker = gatk_docker,
@@ -611,7 +611,7 @@ task M2 {
         gatk --java-options "-Xmx${command_mem}m" GetSampleName -R ${ref_fasta} -I ${tumor_bam} -O tumor_name.txt -encode
         tumor_command_line="-I ${tumor_bam} -tumor `cat tumor_name.txt`"
 
-        if [[ -f "${normal_bam}" ]]; then
+        if [[ ! -z "${normal_bam}" ]]; then
             gatk --java-options "-Xmx${command_mem}m" GetSampleName -R ${ref_fasta} -I ${normal_bam} -O normal_name.txt -encode
             normal_command_line="-I ${normal_bam} -normal `cat normal_name.txt`"
         fi
@@ -634,11 +634,11 @@ task M2 {
         touch tumor-pileups.table
         touch normal-pileups.table
 
-        if [[ -f "${variants_for_contamination}" ]]; then
+        if [[ ! -z "${variants_for_contamination}" ]]; then
             gatk --java-options "-Xmx${command_mem}m" GetPileupSummaries -R ${ref_fasta} -I ${tumor_bam} ${"--interval-set-rule INTERSECTION -L " + intervals} \
                 -V ${variants_for_contamination} -L ${variants_for_contamination} -O tumor-pileups.table
 
-            if [[ -f ${normal_bam} ]]; then
+            if [[ ! -z ${normal_bam} ]]; then
                 gatk --java-options "-Xmx${command_mem}m" GetPileupSummaries -R ${ref_fasta} -I ${normal_bam} ${"--interval-set-rule INTERSECTION -L " + intervals} \
                     -V ${variants_for_contamination} -L ${variants_for_contamination} -O normal-pileups.table
             fi
@@ -915,7 +915,7 @@ task CollectF1R2Counts {
 
     output {
         File alt_table = glob("*-alt.tsv")[0]
-        File ref_histogram = glob("*-ref.metrics")[0]
+        File ref_histograms = glob("*-ref.metrics")[0]
         File alt_histograms = glob("*-alt-depth1.metrics")[0]
         String tumor_sample = read_string("tumor_name.txt")
     }

diff --git a/...rg/broadinstitute/hellbender/tools/walkers/readorientation/LearnReadOrientationModel.java b/...rg/broadinstitute/hellbender/tools/walkers/readorientation/LearnReadOrientationModel.java
@@ -76,12 +76,10 @@ public class LearnReadOrientationModel extends CommandLineProgram {
     protected void onStartup(){
         Utils.validate(refHistogramFiles.size() == altDataTables.size() && altDataTables.size() == altHistogramFiles.size(),
                 "The numbers of ref histograms, alt histograms, and alt table files must all be the same");
-        START HERE CHECK HEADER
-
-        refHistograms = sumHistogramsFromFiles(refHistogramFiles);
+        refHistograms = sumHistogramsFromFiles(refHistogramFiles, true);
 
         if (altHistogramFiles != null) {
-            altHistograms = sumHistogramsFromFiles(altHistogramFiles);
+            altHistograms = sumHistogramsFromFiles(altHistogramFiles, false);
         } else {
             altHistograms = Collections.emptyList();
         }
@@ -258,16 +256,29 @@ public static MetricsFile<?, Integer> readMetricsFile(File file){
         return metricsFile;
     }
 
-    public static List<Histogram<Integer>> sumHistogramsFromFiles(final List<File> files){
+    public static List<Histogram<Integer>> sumHistogramsFromFiles(final List<File> files, final boolean ref){
         Utils.nonNull(files, "files may not be null");
-
         final List<Histogram<Integer>> histogramList = readMetricsFile(files.get(0)).getAllHistograms();
+        if (ref){
+            Utils.validate(histogramList.size() == F1R2FilterConstants.NUM_KMERS,
+                    "The list of ref histograms need to include all kmers as enforced by CollectF1R2Counts");
+            Utils.validate(histogramList.stream().allMatch(h -> F1R2FilterConstants.ALL_KMERS.contains(h.getValueLabel())),
+                    "a histogram contains an unsupported, non-kmer header");
+        } else {
+            Utils.validate(histogramList.size() == F1R2FilterConstants.NUM_KMERS * F1R2FilterConstants.numAltHistogramsPerContext,
+                    "The list of alt histograms missing some (kmer, alt allele, f1r2) triple");
+
+        }
+
+
+
+
         for (int i = 1; i < files.size(); i++){
             final List<Histogram<Integer>> ithHistograms = readMetricsFile(files.get(i)).getAllHistograms();
             for (final Histogram<Integer> jthHistogram : ithHistograms){
                 final String refContext = jthHistogram.getValueLabel();
                 final Optional<Histogram<Integer>> hist = histogramList.stream().filter(h -> h.getValueLabel().equals(refContext)).findAny();
-                Utils.validate(hist.isPresent(),"Reference histogram is empty, which violates the invariant enforced by CollectF1R2Counts");
+                Utils.validate(hist.isPresent(),"Missing histogram header for: " + refContext);
 
                 hist.get().addHistogram(jthHistogram);
             }
@@ -277,7 +288,6 @@ public static List<Histogram<Integer>> sumHistogramsFromFiles(final List<File> f
 
     public static Pair<String, List<AltSiteRecord>> gatherAltSiteRecords(final List<File> tables){
         final int defaultInitialListSize = 1_000_000;
-
         final Pair<String, List<AltSiteRecord>> sampleAndRecords = AltSiteRecord.readAltSiteRecords(tables.get(0), defaultInitialListSize);
         final String sample = sampleAndRecords.getLeft();
 

diff --git a/...institute/hellbender/tools/walkers/readorientation/LearnReadOrientationModelUnitTest.java b/...institute/hellbender/tools/walkers/readorientation/LearnReadOrientationModelUnitTest.java
@@ -55,8 +55,8 @@ public void testCombineHistograms(){
                     CollectF1R2Counts.class.getSimpleName());
         }
 
-        final List<Histogram<Integer>> ref = LearnReadOrientationModel.sumHistogramsFromFiles(Arrays.asList(refMetricsDir.listFiles()));
-        final List<Histogram<Integer>> alt = LearnReadOrientationModel.sumHistogramsFromFiles(Arrays.asList(altMetricsDir.listFiles()));
+        final List<Histogram<Integer>> ref = LearnReadOrientationModel.sumHistogramsFromFiles(Arrays.asList(refMetricsDir.listFiles()), true);
+        final List<Histogram<Integer>> alt = LearnReadOrientationModel.sumHistogramsFromFiles(Arrays.asList(altMetricsDir.listFiles()), false);
         final List<AltSiteRecord> altSites = LearnReadOrientationModel.gatherAltSiteRecords(Arrays.asList(altTableDir.listFiles())).getRight();
 
         final List<Histogram<Integer>> refTruth = LearnReadOrientationModel.readMetricsFile(refHistTruthFile).getAllHistograms();

diff --git a/...stitute/hellbender/tools/walkers/readorientation/ReadOrientationModelIntegrationTest.java b/...stitute/hellbender/tools/walkers/readorientation/ReadOrientationModelIntegrationTest.java
@@ -1,6 +1,5 @@
 package org.broadinstitute.hellbender.tools.walkers.readorientation;
 
-import htsjdk.samtools.util.Histogram;
 import htsjdk.variant.variantcontext.VariantContext;
 import org.apache.commons.lang3.tuple.ImmutableTriple;
 import org.apache.commons.lang3.tuple.Triple;
@@ -13,15 +12,11 @@
 import org.broadinstitute.hellbender.tools.walkers.SplitIntervals;
 import org.broadinstitute.hellbender.tools.walkers.annotator.ReferenceBases;
 import org.broadinstitute.hellbender.tools.walkers.mutect.filtering.FilterMutectCalls;
-import org.broadinstitute.hellbender.tools.walkers.contamination.GatherPileupSummaries;
-import org.broadinstitute.hellbender.tools.walkers.contamination.GetPileupSummaries;
 import org.broadinstitute.hellbender.tools.walkers.mutect.M2ArgumentCollection;
-import org.broadinstitute.hellbender.tools.walkers.mutect.filtering.M2FiltersArgumentCollection;
 import org.broadinstitute.hellbender.tools.walkers.mutect.Mutect2;
 import org.broadinstitute.hellbender.utils.GATKProtectedVariantContextUtils;
 import org.broadinstitute.hellbender.utils.variant.GATKVCFConstants;
 import org.testng.Assert;
-import org.testng.annotations.BeforeClass;
 import org.testng.annotations.DataProvider;
 import org.testng.annotations.Test;
 
@@ -30,7 +25,6 @@
 import java.nio.file.Files;
 import java.nio.file.Paths;
 import java.util.*;
-import java.util.stream.IntStream;
 import java.util.stream.StreamSupport;
 
 public class ReadOrientationModelIntegrationTest extends CommandLineProgramTest {
@@ -85,13 +79,13 @@ public void testOnRealBam(final int scatterCount) throws IOException {
         args.addArgument(StandardArgumentDefinitions.OUTPUT_LONG_NAME, priorTable.getAbsolutePath());
         final File[] refMetricsFiles = refMetricsDir.listFiles();
         Arrays.stream(refMetricsFiles).forEach(f ->
-                args.addArgument(CollectF1R2Counts.REF_SITE_METRICS_SHORT_NAME, f.getAbsolutePath()));
+                args.addArgument(CollectF1R2Counts.REF_SITE_METRICS_LONG_NAME, f.getAbsolutePath()));
         final File[] altMetricsFiles = altMetricsDir.listFiles();
         Arrays.stream(altMetricsFiles).forEach(f ->
-                args.addArgument(CollectF1R2Counts.ALT_DEPTH1_HISTOGRAM_SHORT_NAME, f.getAbsolutePath()));
+                args.addArgument(CollectF1R2Counts.ALT_DEPTH1_HISTOGRAM_LONG_NAME, f.getAbsolutePath()));
         final File[] altTableFiles = altTableDir.listFiles();
         Arrays.stream(altTableFiles).forEach(f ->
-                args.addArgument(CollectF1R2Counts.ALT_DATA_TABLE_SHORT_NAME, f.getAbsolutePath()));
+                args.addArgument(CollectF1R2Counts.ALT_DATA_TABLE_LONG_NAME, f.getAbsolutePath()));
         runCommandLine(args.getArgsList(), LearnReadOrientationModel.class.getSimpleName());
 
         final ArtifactPriorCollection artifactPriorCollection = ArtifactPriorCollection.readArtifactPriors(priorTable);