Package ubic.gemma.core.datastructure.matrix

Class TwoChannelExpressionDataMatrixBuilder

  • public class TwoChannelExpressionDataMatrixBuilder
extends Object
    Utility methods for taking an ExpressionExperiment and returning various types of ExpressionDataMatrices, such as the processed data, preferred data, background, etc. This class is not database aware; use the ExpressionDataMatrixService to get ready-to-use matrices starting from an ExpressionExperiment. This handles complexities such as experiments that contain multiple array designs with differing quantitation types.
    Author:
    pavlidis

    • Method Detail

      • getMissingValueQuantitationTypes

        public static Collection<QuantitationType> getMissingValueQuantitationTypes​(ExpressionExperiment expressionExperiment)
        Parameters:
        expressionExperiment - (should be lightly thawed)
        Returns:
        a collection of QTs

      • getUsefulQuantitationTypes

        public static Collection<QuantitationType> getUsefulQuantitationTypes​(Collection<QuantitationType> eeQtTypes)
        Parameters:
        eeQtTypes - the QTs
        Returns:
        just the quantitation types that are likely to be 'useful': Preferred, present/absent, signals and background from both channels (if present).

      • getUsefulQuantitationTypes

        public static Collection<QuantitationType> getUsefulQuantitationTypes​(ExpressionExperiment expressionExperiment)
        Parameters:
        expressionExperiment - the EE to get the QTs for
        Returns:
        just the quantitation types that are likely to be 'useful': Preferred, present/absent, signals and background from both channels (if present).

      • getIntensity

        public ExpressionDataDoubleMatrix getIntensity()
        Returns:
        Compute an intensity matrix. For two-channel arrays, this is the geometric mean of the background-subtracted signals on the two channels. For two-color arrays, if one channel is missing (as happens sometimes) the intensities returned are just from the one channel. For one-color arrays, this is the same as the preferred data matrix.

      • getMissingValueData

        @Nullable
public ExpressionDataBooleanMatrix getMissingValueData()
        Returns:
        a matrix of booleans, or null if a missing value quantitation type ("absent/present", which may have been computed by our system) is not found. This will return the values whether the array design is two-color or not.

      • getPreferredData

        public ExpressionDataDoubleMatrix getPreferredData()
        Returns:
        The matrix for the preferred data - NOT the processed data (though they may be the same, in fact)

      • isAnyMissing

        public boolean isAnyMissing()

      • logTransformMatrix

        public static void logTransformMatrix​(ExpressionDataDoubleMatrix matrix)
        Log-transform the values in the matrix (base 2). Non-positive values (which have no logarithm defined) are entered as NaN.
        Parameters:
        matrix - matrix

      • addMatrices

        public void addMatrices​(ExpressionDataDoubleMatrix a,
                        ExpressionDataDoubleMatrix b)
        Add two matrices. Ideally, they matrices are conformant, but if they are not (as some rows are sometimes missing for some quantitation types), this method attempts to handle it anyway (see below). The rows and columns do not have to be in the same order, but they do have to have the same column keys and row keys (with the exception of missing rows). The result is stored in a. If the number of rows are not the same, and/or the rows have different keys in the two matrices, some rows will simply not get added and a warning will be issued.
        Parameters:
        a - matrix a
        b - matrix b
        Throws:
        IllegalArgumentException - if the matrices are not column-conformant.