Updated plotting for ModelSegments CNV pipeline. (#3729)

src/main/java/org/broadinstitute/hellbender/tools/copynumber/coverage/copyratio/CopyRatio.java

@@ -0,0 +1,80 @@
+package org.broadinstitute.hellbender.tools.copynumber.coverage.copyratio;
+
+import htsjdk.samtools.util.Locatable;
+import org.broadinstitute.hellbender.utils.SimpleInterval;
+import org.broadinstitute.hellbender.utils.Utils;
+
+public class CopyRatio implements Locatable {
+    private final SimpleInterval interval;
+    private final double log2CopyRatioValue;
+
+    public CopyRatio(final SimpleInterval interval,
+                     final double log2CopyRatioValue) {
+        Utils.nonNull(interval);
+        this.interval = interval;
+        this.log2CopyRatioValue = log2CopyRatioValue;
+    }
+
+    @Override
+    public String getContig() {
+        return interval.getContig();
+    }
+
+    @Override
+    public int getStart() {
+        return interval.getStart();
+    }
+
+    @Override
+    public int getEnd() {
+        return interval.getEnd();
+    }
+
+    public SimpleInterval getInterval() {
+        return interval;
+    }
+
+    public double getLog2CopyRatioValue() {
+        return log2CopyRatioValue;
+    }
+
+    /**
+     * The midpoint is used to characterize the interval for the purposes of determining overlaps when combining
+     * copy-ratio and allele-fraction segmentations, so that each copy-ratio interval will be uniquely contained
+     * in a single segment.
+     */
+    public SimpleInterval getMidpoint() {
+        final int midPoint = (getStart() + getEnd()) / 2;
+        return new SimpleInterval(interval.getContig(), midPoint, midPoint);
+    }
+
+    @Override
+    public boolean equals(Object o) {
+        if (this == o) {
+            return true;
+        }
+        if (o == null || getClass() != o.getClass()) {
+            return false;
+        }
+        final CopyRatio copyRatio = (CopyRatio) o;
+        return Double.compare(copyRatio.log2CopyRatioValue, log2CopyRatioValue) == 0 && interval.equals(copyRatio.interval);
+    }
+
+    @Override
+    public int hashCode() {
+        int result;
+        long temp;
+        result = interval.hashCode();
+        temp = Double.doubleToLongBits(log2CopyRatioValue);
+        result = 31 * result + (int) (temp ^ (temp >>> 32));
+        return result;
+    }
+
+    @Override
+    public String toString() {
+        return "CopyRatio{" +
+                "interval=" + interval +
+                ", log2CopyRatioValue=" + log2CopyRatioValue +
+                '}';
+    }
+}

src/main/java/org/broadinstitute/hellbender/tools/copynumber/coverage/copyratio/CopyRatioCollection.java

@@ -0,0 +1,64 @@
+package org.broadinstitute.hellbender.tools.copynumber.coverage.copyratio;
+
+import htsjdk.samtools.util.OverlapDetector;
+import org.broadinstitute.hellbender.tools.copynumber.formats.collections.SampleLocatableCollection;
+import org.broadinstitute.hellbender.tools.copynumber.formats.metadata.SampleMetadata;
+import org.broadinstitute.hellbender.utils.SimpleInterval;
+import org.broadinstitute.hellbender.utils.tsv.DataLine;
+import org.broadinstitute.hellbender.utils.tsv.TableColumnCollection;
+
+import java.io.File;
+import java.util.List;
+import java.util.function.BiConsumer;
+import java.util.function.Function;
+import java.util.stream.Collectors;
+
+public final class CopyRatioCollection extends SampleLocatableCollection<CopyRatio> {
+    enum CopyRatioTableColumn {
+        CONTIG,
+        START,
+        END,
+        LOG2_COPY_RATIO;
+
+        static final TableColumnCollection COLUMNS = new TableColumnCollection((Object[]) values());
+    }
+
+    private static final Function<DataLine, CopyRatio> COPY_RATIO_RECORD_FROM_DATA_LINE_DECODER = dataLine -> {
+        final String contig = dataLine.get(CopyRatioTableColumn.CONTIG);
+        final int start = dataLine.getInt(CopyRatioTableColumn.START);
+        final int end = dataLine.getInt(CopyRatioTableColumn.END);
+        final double copyRatio = dataLine.getDouble(CopyRatioTableColumn.LOG2_COPY_RATIO);
+        final SimpleInterval interval = new SimpleInterval(contig, start, end);
+        return new CopyRatio(interval, copyRatio);
+    };
+
+    private static final BiConsumer<CopyRatio, DataLine> COPY_RATIO_RECORD_TO_DATA_LINE_ENCODER = (copyRatio, dataLine) ->
+        dataLine.append(copyRatio.getInterval().getContig())
+                .append(copyRatio.getInterval().getStart())
+                .append(copyRatio.getInterval().getEnd())
+                .append(copyRatio.getLog2CopyRatioValue());
+
+    public CopyRatioCollection(final File inputFile) {
+        super(inputFile, CopyRatioTableColumn.COLUMNS, COPY_RATIO_RECORD_FROM_DATA_LINE_DECODER, COPY_RATIO_RECORD_TO_DATA_LINE_ENCODER);
+    }
+
+    public CopyRatioCollection(final SampleMetadata sampleMetadata,
+                               final List<CopyRatio> copyRatios) {
+        super(sampleMetadata, copyRatios, CopyRatioTableColumn.COLUMNS, COPY_RATIO_RECORD_FROM_DATA_LINE_DECODER, COPY_RATIO_RECORD_TO_DATA_LINE_ENCODER);
+    }
+
+    public List<Double> getLog2CopyRatioValues() {
+        return getRecords().stream().map(CopyRatio::getLog2CopyRatioValue).collect(Collectors.toList());
+    }
+
+    /**
+     * The midpoint is used to characterize the interval for the purposes of determining overlaps when combining
+     * copy-ratio and allele-fraction segmentations, so that each copy-ratio interval will be uniquely contained
+     * in a single segment.
+     */
+    public OverlapDetector<CopyRatio> getMidpointOverlapDetector() {
+        return OverlapDetector.create(getRecords().stream()
+                .map(cr -> new CopyRatio(cr.getMidpoint(), cr.getLog2CopyRatioValue()))
+                .collect(Collectors.toList()));
+    }
+}

src/main/java/org/broadinstitute/hellbender/tools/copynumber/formats/CopyNumberStandardArgument.java

@@ -0,0 +1,30 @@
+package org.broadinstitute.hellbender.tools.copynumber.formats;
+
+public final class CopyNumberStandardArgument {
+    public static final String ANNOTATED_INTERVALS_FILE_LONG_NAME = "annotatedIntervals";
+    public static final String ANNOTATED_INTERVALS_FILE_SHORT_NAME = "annot";
+
+    public static final String READ_COUNT_PANEL_OF_NORMALS_FILE_LONG_NAME = "readCountPanelOfNormals";
+    public static final String READ_COUNT_PANEL_OF_NORMALS_FILE_SHORT_NAME = "rcPON";
+
+    public static final String NUMBER_OF_EIGENSAMPLES_LONG_NAME = "numberOfEigensamples";
+    public static final String NUMBER_OF_EIGENSAMPLES_SHORT_NAME = "numEigen";
+
+    public static final String STANDARDIZED_COPY_RATIOS_FILE_LONG_NAME = "standardizedCopyRatios";
+    public static final String STANDARDIZED_COPY_RATIOS_FILE_SHORT_NAME = "standardizedCR";
+
+    public static final String DENOISED_COPY_RATIOS_FILE_LONG_NAME = "denoisedCopyRatios";
+    public static final String DENOISED_COPY_RATIOS_FILE_SHORT_NAME = "denoisedCR";
+
+    public static final String ALLELIC_COUNTS_FILE_LONG_NAME = "allelicCounts";
+    public static final String ALLELIC_COUNTS_FILE_SHORT_NAME = "AC";
+
+    public static final String NORMAL_ALLELIC_COUNTS_FILE_LONG_NAME = "normalAllelicCounts";
+    public static final String NORMAL_ALLELIC_COUNTS_FILE_SHORT_NAME = "normalAC";
+
+    public static final String SEGMENTS_FILE_LONG_NAME = "segments";
+    public static final String SEGMENTS_FILE_SHORT_NAME = "S";
+
+    public static final String OUTPUT_PREFIX_LONG_NAME = "outputPrefix";
+    public static final String OUTPUT_PREFIX_SHORT_NAME = "pre";
+}

src/main/java/org/broadinstitute/hellbender/tools/copynumber/formats/collections/SampleLocatableCollection.java

@@ -5,6 +5,7 @@
 import htsjdk.samtools.util.OverlapDetector;
 import org.broadinstitute.hellbender.tools.copynumber.formats.metadata.SampleMetadata;
 import org.broadinstitute.hellbender.utils.IntervalUtils;
+import org.broadinstitute.hellbender.utils.SimpleInterval;
 import org.broadinstitute.hellbender.utils.Utils;
 import org.broadinstitute.hellbender.utils.tsv.DataLine;
 import org.broadinstitute.hellbender.utils.tsv.TableColumnCollection;
@@ -76,6 +77,16 @@ private static <T extends Locatable> void validateIntervals(final String sampleN
         }
     }
 
+    /**
+     * @return  a new modifiable list of {@link SimpleInterval}s corresponding to the {@link Locatable}s
+     *          for each record contained in the collection
+     */
+    public List<SimpleInterval> getIntervals() {
+        return getRecords().stream()
+                .map(r -> new SimpleInterval(r.getContig(), r.getStart(), r.getEnd()))
+                .collect(Collectors.toList());
+    }
+
     public OverlapDetector<RECORD> getOverlapDetector() {
         return OverlapDetector.create(getRecords());
     }

src/main/java/org/broadinstitute/hellbender/tools/copynumber/multidimensional/model/ModeledSegment.java

@@ -0,0 +1,109 @@
+package org.broadinstitute.hellbender.tools.copynumber.multidimensional.model;
+
+import htsjdk.samtools.util.Locatable;
+import org.broadinstitute.hellbender.utils.SimpleInterval;
+import org.broadinstitute.hellbender.utils.Utils;
+import org.broadinstitute.hellbender.utils.mcmc.Decile;
+import org.broadinstitute.hellbender.utils.mcmc.DecileCollection;
+
+import java.util.List;
+
+public class ModeledSegment implements Locatable {
+    private final SimpleInterval interval;
+    private final int numPointsCopyRatio;
+    private final int numPointsAlleleFraction;
+
+    private final SimplePosteriorSummary log2CopyRatioSimplePosteriorSummary;
+    private final SimplePosteriorSummary minorAlleleFractionSimplePosteriorSummary;
+
+    public ModeledSegment(final SimpleInterval interval,
+                          final int numPointsCopyRatio,
+                          final int numPointsAlleleFraction,
+                          final SimplePosteriorSummary log2CopyRatioSimplePosteriorSummary,
+                          final SimplePosteriorSummary minorAlleleFractionSimplePosteriorSummary) {
+        Utils.validateArg(numPointsCopyRatio > 0 || numPointsAlleleFraction > 0,
+                String.format("Number of copy-ratio points or number of allele-fraction points must be positive: %s", interval));
+        this.interval = Utils.nonNull(interval);
+        this.numPointsCopyRatio = numPointsCopyRatio;
+        this.numPointsAlleleFraction = numPointsAlleleFraction;
+        this.log2CopyRatioSimplePosteriorSummary = Utils.nonNull(log2CopyRatioSimplePosteriorSummary);
+        this.minorAlleleFractionSimplePosteriorSummary = Utils.nonNull(minorAlleleFractionSimplePosteriorSummary);
+    }
+
+    @Override
+    public String getContig() {
+        return interval.getContig();
+    }
+
+    @Override
+    public int getStart() {
+        return interval.getStart();
+    }
+
+    @Override
+    public int getEnd() {
+        return interval.getEnd();
+    }
+
+    public SimpleInterval getInterval() {
+        return interval;
+    }
+
+    public int getNumPointsCopyRatio() {
+        return numPointsCopyRatio;
+    }
+
+    public int getNumPointsAlleleFraction() {
+        return numPointsAlleleFraction;
+    }
+
+    public SimplePosteriorSummary getLog2CopyRatioSimplePosteriorSummary() {
+        return log2CopyRatioSimplePosteriorSummary;
+    }
+
+    public SimplePosteriorSummary getMinorAlleleFractionSimplePosteriorSummary() {
+        return minorAlleleFractionSimplePosteriorSummary;
+    }
+
+    public static final class SimplePosteriorSummary {
+        private final double decile10;
+        private final double decile50;
+        private final double decile90;
+
+        public SimplePosteriorSummary(final double decile10,
+                                      final double decile50,
+                                      final double decile90) {
+            this.decile10 = decile10;
+            this.decile50 = decile50;
+            this.decile90 = decile90;
+        }
+
+        public SimplePosteriorSummary(final List<Double> samples) {
+            final DecileCollection deciles = new DecileCollection(samples);
+            this.decile10 = deciles.get(Decile.DECILE_10);
+            this.decile50 = deciles.get(Decile.DECILE_50);
+            this.decile90 = deciles.get(Decile.DECILE_90);
+        }
+
+        public double getDecile10() {
+            return decile10;
+        }
+
+        public double getDecile50() {
+            return decile50;
+        }
+
+        public double getDecile90() {
+            return decile90;
+        }
+
+        @Override
+        public String toString() {
+            return "SimplePosteriorSummary{" +
+                    "decile10=" + decile10 +
+                    ", decile50=" + decile50 +
+                    ", decile90=" + decile90 +
+                    '}';
+        }
+    }
+}

src/main/java/org/broadinstitute/hellbender/tools/copynumber/multidimensional/model/ModeledSegmentCollection.java

@@ -0,0 +1,75 @@
+package org.broadinstitute.hellbender.tools.copynumber.multidimensional.model;
+
+import org.broadinstitute.hellbender.tools.copynumber.formats.collections.SampleLocatableCollection;
+import org.broadinstitute.hellbender.tools.copynumber.formats.metadata.SampleMetadata;
+import org.broadinstitute.hellbender.utils.SimpleInterval;
+import org.broadinstitute.hellbender.utils.tsv.DataLine;
+import org.broadinstitute.hellbender.utils.tsv.TableColumnCollection;
+
+import java.io.File;
+import java.util.List;
+import java.util.function.BiConsumer;
+import java.util.function.Function;
+
+/**
+ * @author Samuel Lee <slee@broadinstitute.org>
+ */
+public final class ModeledSegmentCollection extends SampleLocatableCollection<ModeledSegment> {
+    private static final String DOUBLE_FORMAT = "%6.6f";    //TODO replace this with MultidimensionalModeller.DOUBLE_FORMAT from sl_wgs_acnv branch
+
+    enum ModeledSegmentTableColumn {
+        CONTIG, 
+        START, 
+        END,
+        NUM_POINTS_COPY_RATIO,
+        NUM_POINTS_ALLELE_FRACTION,
+        LOG2_COPY_RATIO_POSTERIOR_10,
+        LOG2_COPY_RATIO_POSTERIOR_50,
+        LOG2_COPY_RATIO_POSTERIOR_90,
+        MINOR_ALLELE_FRACTION_POSTERIOR_10,
+        MINOR_ALLELE_FRACTION_POSTERIOR_50,
+        MINOR_ALLELE_FRACTION_POSTERIOR_90;
+
+        static final TableColumnCollection COLUMNS = new TableColumnCollection((Object[]) values());
+    }
+
+    private static final Function<DataLine, ModeledSegment> MODELED_SEGMENT_RECORD_FROM_DATA_LINE_DECODER = dataLine -> {
+        final String contig = dataLine.get(ModeledSegmentTableColumn.CONTIG);
+        final int start = dataLine.getInt(ModeledSegmentTableColumn.START);
+        final int end = dataLine.getInt(ModeledSegmentTableColumn.END);
+        final int numPointsCopyRatio = dataLine.getInt(ModeledSegmentTableColumn.NUM_POINTS_COPY_RATIO);
+        final int numPointsAlleleFraction = dataLine.getInt(ModeledSegmentTableColumn.NUM_POINTS_ALLELE_FRACTION);
+        final double log2CopyRatioPosterior10 = dataLine.getDouble(ModeledSegmentTableColumn.LOG2_COPY_RATIO_POSTERIOR_10);
+        final double log2CopyRatioPosterior50 = dataLine.getDouble(ModeledSegmentTableColumn.LOG2_COPY_RATIO_POSTERIOR_50);
+        final double log2CopyRatioPosterior90 = dataLine.getDouble(ModeledSegmentTableColumn.LOG2_COPY_RATIO_POSTERIOR_90);
+        final double minorAlleleFractionPosterior10 = dataLine.getDouble(ModeledSegmentTableColumn.MINOR_ALLELE_FRACTION_POSTERIOR_10);
+        final double minorAlleleFractionPosterior50 = dataLine.getDouble(ModeledSegmentTableColumn.MINOR_ALLELE_FRACTION_POSTERIOR_50);
+        final double minorAlleleFractionPosterior90 = dataLine.getDouble(ModeledSegmentTableColumn.MINOR_ALLELE_FRACTION_POSTERIOR_90);
+        final SimpleInterval interval = new SimpleInterval(contig, start, end);
+        return new ModeledSegment(interval, numPointsCopyRatio, numPointsAlleleFraction, 
+                new ModeledSegment.SimplePosteriorSummary(log2CopyRatioPosterior10, log2CopyRatioPosterior50, log2CopyRatioPosterior90),
+                new ModeledSegment.SimplePosteriorSummary(minorAlleleFractionPosterior10, minorAlleleFractionPosterior50, minorAlleleFractionPosterior90));
+    };
+
+    private static final BiConsumer<ModeledSegment, DataLine> MODELED_SEGMENT_RECORD_TO_DATA_LINE_ENCODER = (modeledSegment, dataLine) ->
+            dataLine.append(modeledSegment.getContig())
+                    .append(modeledSegment.getStart())
+                    .append(modeledSegment.getEnd())
+                    .append(modeledSegment.getNumPointsCopyRatio())
+                    .append(modeledSegment.getNumPointsAlleleFraction())
+                    .append(String.format(DOUBLE_FORMAT, modeledSegment.getLog2CopyRatioSimplePosteriorSummary().getDecile10()))
+                    .append(String.format(DOUBLE_FORMAT, modeledSegment.getLog2CopyRatioSimplePosteriorSummary().getDecile50()))
+                    .append(String.format(DOUBLE_FORMAT, modeledSegment.getLog2CopyRatioSimplePosteriorSummary().getDecile90()))
+                    .append(String.format(DOUBLE_FORMAT, modeledSegment.getMinorAlleleFractionSimplePosteriorSummary().getDecile10()))
+                    .append(String.format(DOUBLE_FORMAT, modeledSegment.getMinorAlleleFractionSimplePosteriorSummary().getDecile50()))
+                    .append(String.format(DOUBLE_FORMAT, modeledSegment.getMinorAlleleFractionSimplePosteriorSummary().getDecile90()));
+
+    public ModeledSegmentCollection(final File inputFile) {
+        super(inputFile, ModeledSegmentTableColumn.COLUMNS, MODELED_SEGMENT_RECORD_FROM_DATA_LINE_DECODER, MODELED_SEGMENT_RECORD_TO_DATA_LINE_ENCODER);
+    }
+
+    public ModeledSegmentCollection(final SampleMetadata sampleMetadata,
+                                    final List<ModeledSegment> modeledSegments) {
+        super(sampleMetadata, modeledSegments, ModeledSegmentTableColumn.COLUMNS, MODELED_SEGMENT_RECORD_FROM_DATA_LINE_DECODER, MODELED_SEGMENT_RECORD_TO_DATA_LINE_ENCODER);
+    }
+}