org.broadinstitute.hellbender.tools.copynumber.denoising

Class HDF5SVDReadCountPanelOfNormals

java.lang.Object

org.broadinstitute.hellbender.tools.copynumber.denoising.HDF5SVDReadCountPanelOfNormals

All Implemented Interfaces:

SVDReadCountPanelOfNormals

public final class HDF5SVDReadCountPanelOfNormals
extends java.lang.Object
implements SVDReadCountPanelOfNormals

Represents the SVD panel of normals to be created by CreateReadCountPanelOfNormals.

Data is stored in the following HDF5 paths:

• version: /version/value
• command line: /command_line/value
• sequence dictionary: /sequence_dictionary/value
• original read counts: /original_data/read_counts_samples_by_intervals
• original sample filenames: /original_data/sample_filenames
• original intervals: /original_data/intervals
• interval GC content (optional): /original_data/interval_gc_content
• panel sample filenames (after filtering): /panel/sample_filenames
• panel intervals (after filtering): /panel/intervals
• panel interval fractional medians: /panel/interval_fractional_medians
• panel singular values: /panel/singular_values
• panel interval fractional medians: /panel/interval_fractional_medians
• panel eigensamples: /panel/transposed_eigensamples_samples_by_intervals

Most attributes are stored as wide matrices (i.e., more columns than rows) when possible. This avoids a very slow write time in HDF5, since HDF5 writes wide matrices much faster than tall matrices. See HDF5Utils.writeIntervals(org.broadinstitute.hdf5.HDF5File, java.lang.String, java.util.List<T>) for details on the representation of intervals.

Method Summary

Modifier and Type	Method and Description
static void	create(java.io.File outFile, java.lang.String commandLine, htsjdk.samtools.SAMSequenceDictionary sequenceDictionary, org.apache.commons.math3.linear.RealMatrix originalReadCounts, java.util.List<java.lang.String> originalSampleFilenames, java.util.List<SimpleInterval> originalIntervals, double[] intervalGCContent, double minimumIntervalMedianPercentile, double maximumZerosInSamplePercentage, double maximumZerosInIntervalPercentage, double extremeSampleMedianPercentile, boolean doImputeZeros, double extremeOutlierTruncationPercentile, int numEigensamplesRequested, int maximumChunkSize, org.apache.spark.api.java.JavaSparkContext ctx) Create the panel of normals and write it to an HDF5 file.
double[][]	getEigensampleVectors() Returns a modifiable copy of an array containing the orthnonormal matrix of eigensample vectors.
int	getNumEigensamples() Returns the number of eigensamples.
double[]	getOriginalIntervalGCContent() Returns a modifiable copy of an array containing the GC content of the original intervals (in the same order as in SVDReadCountPanelOfNormals.getOriginalIntervals()).
java.util.List<SimpleInterval>	getOriginalIntervals() Returns a modifiable copy of the list of the original intervals that were used to build this PoN (no filtering will have been applied).
double[][]	getOriginalReadCounts() Returns a modifiable copy of the original matrix of integer read-counts (represented as doubles) used to build the PoN (no filtering will have been applied).
double[]	getPanelIntervalFractionalMedians() Returns a modifiable copy of an array containing the median (across all samples, before filtering) of the fractional coverage at each panel interval (in the same order as in SVDReadCountPanelOfNormals.getPanelIntervals()).
java.util.List<SimpleInterval>	getPanelIntervals() Returns a modifiable copy of the list of the intervals contained in this PoN after all filtering has been applied.
htsjdk.samtools.SAMSequenceDictionary	getSequenceDictionary() Returns the sequence dictionary common to all of the read counts used to build the PoN.
double[]	getSingularValues() Returns a modifiable copy of an array of the singular values of the eigensamples in decreasing order.
double	getVersion() Returns the PoN version.
static HDF5SVDReadCountPanelOfNormals	read(org.broadinstitute.hdf5.HDF5File file) Create an interface to an HDF5 file.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Methods inherited from interface org.broadinstitute.hellbender.tools.copynumber.denoising.SVDReadCountPanelOfNormals

denoise

Method Detail

getVersion

public double getVersion()

Description copied from interface: SVDReadCountPanelOfNormals

Returns the PoN version.

Specified by:

getVersion in interface SVDReadCountPanelOfNormals

getNumEigensamples

public int getNumEigensamples()

Description copied from interface: SVDReadCountPanelOfNormals

Returns the number of eigensamples.

Specified by:

getNumEigensamples in interface SVDReadCountPanelOfNormals

getSequenceDictionary

public htsjdk.samtools.SAMSequenceDictionary getSequenceDictionary()

Description copied from interface: SVDReadCountPanelOfNormals

Returns the sequence dictionary common to all of the read counts used to build the PoN.

Specified by:

getSequenceDictionary in interface SVDReadCountPanelOfNormals

getOriginalReadCounts

public double[][] getOriginalReadCounts()

Description copied from interface: SVDReadCountPanelOfNormals

Returns a modifiable copy of the original matrix of integer read-counts (represented as doubles) used to build the PoN (no filtering will have been applied). This matrix has has dimensions M_original x N_original, where M_original is the number of original intervals and N_original is the number of original samples.

Specified by:

getOriginalReadCounts in interface SVDReadCountPanelOfNormals

getOriginalIntervals

public java.util.List<SimpleInterval> getOriginalIntervals()

Description copied from interface: SVDReadCountPanelOfNormals

Returns a modifiable copy of the list of the original intervals that were used to build this PoN (no filtering will have been applied). This list has length M_original.

Specified by:

getOriginalIntervals in interface SVDReadCountPanelOfNormals

getOriginalIntervalGCContent

public double[] getOriginalIntervalGCContent()

Description copied from interface: SVDReadCountPanelOfNormals

Returns a modifiable copy of an array containing the GC content of the original intervals (in the same order as in SVDReadCountPanelOfNormals.getOriginalIntervals()). This array has length M_original.

Specified by:

getOriginalIntervalGCContent in interface SVDReadCountPanelOfNormals

getPanelIntervals

public java.util.List<SimpleInterval> getPanelIntervals()

Description copied from interface: SVDReadCountPanelOfNormals

Returns a modifiable copy of the list of the intervals contained in this PoN after all filtering has been applied. This list has length M.

Specified by:

getPanelIntervals in interface SVDReadCountPanelOfNormals

getPanelIntervalFractionalMedians

public double[] getPanelIntervalFractionalMedians()

Description copied from interface: SVDReadCountPanelOfNormals

Returns a modifiable copy of an array containing the median (across all samples, before filtering) of the fractional coverage at each panel interval (in the same order as in SVDReadCountPanelOfNormals.getPanelIntervals()). This is used to standardize samples. This array has length M.

Specified by:

getPanelIntervalFractionalMedians in interface SVDReadCountPanelOfNormals

getSingularValues

public double[] getSingularValues()

Description copied from interface: SVDReadCountPanelOfNormals

Returns a modifiable copy of an array of the singular values of the eigensamples in decreasing order. This array has length K.

Specified by:

getSingularValues in interface SVDReadCountPanelOfNormals

getEigensampleVectors

public double[][] getEigensampleVectors()

Description copied from interface: SVDReadCountPanelOfNormals

Returns a modifiable copy of an array containing the orthnonormal matrix of eigensample vectors. This matrix has has dimensions M x K, where M is the number of panel intervals (after filtering) and K is the number of eigensamples. Columns are sorted by singular value in decreasing order.

Specified by:

getEigensampleVectors in interface SVDReadCountPanelOfNormals

read

public static HDF5SVDReadCountPanelOfNormals read(org.broadinstitute.hdf5.HDF5File file)

Create an interface to an HDF5 file. A version check is performed and a warning message logged if the version number is not up to date.

create

public static void create(java.io.File outFile,
                          java.lang.String commandLine,
                          htsjdk.samtools.SAMSequenceDictionary sequenceDictionary,
                          org.apache.commons.math3.linear.RealMatrix originalReadCounts,
                          java.util.List<java.lang.String> originalSampleFilenames,
                          java.util.List<SimpleInterval> originalIntervals,
                          double[] intervalGCContent,
                          double minimumIntervalMedianPercentile,
                          double maximumZerosInSamplePercentage,
                          double maximumZerosInIntervalPercentage,
                          double extremeSampleMedianPercentile,
                          boolean doImputeZeros,
                          double extremeOutlierTruncationPercentile,
                          int numEigensamplesRequested,
                          int maximumChunkSize,
                          org.apache.spark.api.java.JavaSparkContext ctx)

Create the panel of normals and write it to an HDF5 file. All inputs are assumed to be valid. The dimensions of originalReadCounts should be samples x intervals. To reduce memory footprint, originalReadCounts is modified in place. If intervalGCContent is null, GC-bias correction will not be performed.