Class DiffExAnalyzerUtils

java.lang.Object

ubic.gemma.core.analysis.expression.diff.DiffExAnalyzerUtils

public class DiffExAnalyzerUtils
extends Object

Field Summary

Fields

Modifier and Type	Field	Description
static final String	BIO_MATERIAL_RNAME_PREFIX
static final String	FACTOR_RNAME_PREFIX
static final String	FACTOR_VALUE_BASELINE_SUFFIX
static final String	FACTOR_VALUE_RNAME_PREFIX

Constructor Summary

Constructors

Constructor	Description
DiffExAnalyzerUtils()

Method Summary

Modifier and Type	Method	Description
static ObjectMatrix<BioMaterial, ExperimentalFactor, Object>	buildDesignMatrix(List<ExperimentalFactor> factors, List<BioMaterial> samplesUsed, boolean allowMissingValues)	Build a design matrix for the given factors and samples.
static ObjectMatrix<String,String,Object>	buildRDesignMatrix(List<ExperimentalFactor> factors, List<BioMaterial> samplesUsed, boolean allowMissingValues)	Build an R-friendly design matrix.
static ObjectMatrix<String,String,Object>	buildRDesignMatrix(List<ExperimentalFactor> factors, List<BioMaterial> samplesUsed, Map<ExperimentalFactor, FactorValue> baselines, boolean allowMissingValues)	A variant of buildRDesignMatrix(List, List, boolean) that allows for reusing baselines for repeated calls.
static BioAssayDimension	createBADMap(List<BioMaterial> columnsToUse)
static Protocol	createProtocolForConfig(DifferentialExpressionAnalysisConfig config, Map<ExperimentalFactor, FactorValue> baselineFactorValues)
static AnalysisType	determineAnalysisType(BioAssaySet bioAssaySet, Collection<ExperimentalFactor> experimentalFactors, ExperimentalFactor subsetFactor, boolean includeInteractionsIfPossible)	Determines the analysis to execute based on the experimental factors, factor values, and block design.
static AnalysisType	determineAnalysisType(BioAssaySet bioAssaySet, Collection<ExperimentalFactor> experimentalFactors, ExperimentalFactor subsetFactor, boolean includeInteractionsIfPossible, Collection<BioMaterial> samplesToAnalyze)	A variant that decides from a given set of samples rather than every sample in the experiment.
static AnalysisType	determineAnalysisType(BioAssaySet bioAssaySet, DifferentialExpressionAnalysisConfig config)
static AnalysisType	determineAnalysisType(BioAssaySet bioAssaySet, DifferentialExpressionAnalysisConfig config, Collection<BioMaterial> samplesToAnalyze)	A variant that decides from a given set of samples rather than every sample in the experiment.
static ExpressionDataDoubleMatrix	dropSamplesNotAnalyzed(ExpressionDataDoubleMatrix dmatrix)	Remove the samples that are not analyzed (see isAnalyzed(BioMaterial)) from a data matrix.
static String	formatInteraction(Set<ExperimentalFactor> i)	Format an interaction of factors.
static boolean	isAnalyzed(BioMaterial sample)	Check whether a sample takes part in a differential expression analysis.
static DoubleMatrix<String,String>	makeDataMatrix(ObjectMatrix<String,String,Object> designMatrix, DoubleMatrix<CompositeSequence, BioMaterial> namedMatrix)	Convert the data into a string-keyed matrix.
static String	nameForR(BioMaterial sample)	Create a name for a sample suitable for R.
static String	nameForR(CompositeSequence cs)
static String	nameForR(ExperimentalFactor experimentalFactor)	Create a name for the factor that is suitable for R.
static String	nameForR(FactorValue fv, boolean isBaseline)	Create a name for the factor value that is suitable for R.
static void	populateFactorValues(Collection<BioMaterial> samples, ExperimentalFactor f, Collection<FactorValue> fvs)	As above, over a given set of samples rather than every sample of the experiment or subset.
static void	populateFactorValuesFromBASet(BioAssaySet ee, ExperimentalFactor f, Collection<FactorValue> fvs)

Methods inherited from class Object

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

Field Details

BIO_MATERIAL_RNAME_PREFIX

public static final String BIO_MATERIAL_RNAME_PREFIX

See Also:

• Constant Field Values

FACTOR_RNAME_PREFIX

public static final String FACTOR_RNAME_PREFIX

See Also:

• Constant Field Values

FACTOR_VALUE_RNAME_PREFIX

public static final String FACTOR_VALUE_RNAME_PREFIX

See Also:

• Constant Field Values

FACTOR_VALUE_BASELINE_SUFFIX

public static final String FACTOR_VALUE_BASELINE_SUFFIX

See Also:

• Constant Field Values

Constructor Details

DiffExAnalyzerUtils

public DiffExAnalyzerUtils()

Method Details

isAnalyzed

public static boolean isAnalyzed(BioMaterial sample)

Check whether a sample takes part in a differential expression analysis.

Two kinds of sample do not: those a curator has marked DE_Exclude, and those whose assays are all flagged as outliers. Both must be removed before anything is decided from the sample set — baselines, whether a factor is complete, the analysis type and the degrees of freedom are all computed from it.

dropSamplesNotAnalyzed

public static ExpressionDataDoubleMatrix dropSamplesNotAnalyzed(ExpressionDataDoubleMatrix dmatrix)

Remove the samples that are not analyzed (see isAnalyzed(BioMaterial)) from a data matrix.

The matrix is returned unchanged when there is nothing to drop, which is the common case.

Throws:

IllegalStateException - if every sample would be dropped

createBADMap

public static BioAssayDimension createBADMap(List<BioMaterial> columnsToUse)

populateFactorValuesFromBASet

public static void populateFactorValuesFromBASet(BioAssaySet ee,
 ExperimentalFactor f,
 Collection<FactorValue> fvs)

populateFactorValues

public static void populateFactorValues(Collection<BioMaterial> samples,
 ExperimentalFactor f,
 Collection<FactorValue> fvs)

As above, over a given set of samples rather than every sample of the experiment or subset. Counting levels over the raw set lets a factor whose second level is carried only by an excluded or outlier sample look variable, so it survives into the model as a constant column.

makeDataMatrix

public static DoubleMatrix<String,String> makeDataMatrix(ObjectMatrix<String,String,Object> designMatrix,
 DoubleMatrix<CompositeSequence, BioMaterial> namedMatrix)

Convert the data into a string-keyed matrix. Assumes that the row names of the designMatrix are concordant with the column names of the namedMatrix

nameForR

public static String nameForR(CompositeSequence cs)

determineAnalysisType

public static AnalysisType determineAnalysisType(BioAssaySet bioAssaySet,
 DifferentialExpressionAnalysisConfig config)

Returns:

selected type of analysis such as t-test, two-way ANOVA, etc.

determineAnalysisType

public static AnalysisType determineAnalysisType(BioAssaySet bioAssaySet,
 DifferentialExpressionAnalysisConfig config,
 @Nullable
 Collection<BioMaterial> samplesToAnalyze)

A variant that decides from a given set of samples rather than every sample in the experiment.

Parameters:

samplesToAnalyze - the samples that will actually be modelled, or null to use all of them. Pass the samples left after dropSamplesNotAnalyzed(ExpressionDataDoubleMatrix), so that DE_Exclude and outlier samples do not decide the analysis type.

Returns:

selected type of analysis such as t-test, two-way ANOVA, etc.

determineAnalysisType

@Nullable
public static AnalysisType determineAnalysisType(BioAssaySet bioAssaySet,
 Collection<ExperimentalFactor> experimentalFactors,
 @Nullable
 ExperimentalFactor subsetFactor,
 boolean includeInteractionsIfPossible)

Determines the analysis to execute based on the experimental factors, factor values, and block design.

This overload judges from every sample in the experiment. Prefer the variant taking samplesToAnalyze when DE_Exclude or outlier samples may be present — see dropSamplesNotAnalyzed(ExpressionDataDoubleMatrix).

Parameters:

bioAssaySet - experiment or subset to determine the analysis type for

experimentalFactors - which factors to use, or null if to use all from the experiment

subsetFactor - can be null

includeInteractionsIfPossible - include interactions among the provided experimental factors if possible

Returns:

an appropriate analysis type

determineAnalysisType

public static AnalysisType determineAnalysisType(BioAssaySet bioAssaySet,
 Collection<ExperimentalFactor> experimentalFactors,
 @Nullable
 ExperimentalFactor subsetFactor,
 boolean includeInteractionsIfPossible,
 @Nullable
 Collection<BioMaterial> samplesToAnalyze)

A variant that decides from a given set of samples rather than every sample in the experiment.

Parameters:

samplesToAnalyze - the samples that will actually be modelled, or null to use all of them

createProtocolForConfig

public static Protocol createProtocolForConfig(DifferentialExpressionAnalysisConfig config,
 Map<ExperimentalFactor, FactorValue> baselineFactorValues)

formatInteraction

public static String formatInteraction(Set<ExperimentalFactor> i)

Format an interaction of factors.

buildDesignMatrix

public static ObjectMatrix<BioMaterial, ExperimentalFactor, Object> buildDesignMatrix(List<ExperimentalFactor> factors,
 List<BioMaterial> samplesUsed,
 boolean allowMissingValues)

Build a design matrix for the given factors and samples.

Parameters:

factors - factors

samplesUsed - the samples used

allowMissingValues - whether to allow missing values, if set to true, the returned matrix may contain nulls

Returns:

the experimental design matrix

Throws:

IllegalStateException - if missing values are found and allowMissingValues is false

buildRDesignMatrix

public static ObjectMatrix<String,String,Object> buildRDesignMatrix(List<ExperimentalFactor> factors,
 List<BioMaterial> samplesUsed,
 boolean allowMissingValues)

Build an R-friendly design matrix.

Rows and columns use names derived from nameForR(BioMaterial), nameForR(ExperimentalFactor) and nameForR(FactorValue, boolean) such that the resulting matrix can be passed to R for analysis. It is otherwise identical to buildDesignMatrix(List, List, boolean).

buildRDesignMatrix

public static ObjectMatrix<String,String,Object> buildRDesignMatrix(List<ExperimentalFactor> factors,
 List<BioMaterial> samplesUsed,
 Map<ExperimentalFactor, FactorValue> baselines,
 boolean allowMissingValues)

A variant of buildRDesignMatrix(List, List, boolean) that allows for reusing baselines for repeated calls. This is used for subset analysis.

nameForR

public static String nameForR(BioMaterial sample)

Create a name for a sample suitable for R.

nameForR

public static String nameForR(ExperimentalFactor experimentalFactor)

Create a name for the factor that is suitable for R.

nameForR

public static String nameForR(FactorValue fv,
 boolean isBaseline)

Create a name for the factor value that is suitable for R.