Eigensystem-class: The Eigensystem class

In biosvd: Package for high-throughput data processing, outlier detection, noise removal and dynamic modeling

Description

Eigensystem is a list-based class for storing the results of applying Singular Value Decomposition (SVD) to a feature by assay data set. Objects are normally created by compute,Eigensystem-method.

Details

The Eigensystem class stores the original data and all SVD-derived information obtained with compute. Data in the Eigensystem are organized into different slots, ‘matrix’, ‘signMatrix’, ‘assayMatrix’, ‘featureMatrix’, ‘eigenassays’, ‘eigenexpressions’, ‘eigenfeatures’, ‘assaycorrelations’, ‘featurecorrelations’, ‘fractions’, ‘entropy’, ‘apply’, ‘excludeEigenfeatures’, and ‘colorIdFeatures’. Brief descriptions of these slots are provided below.

Slots

Eigensystem objects contain the following slots

matrix:

matrix containing the feature by assay data without missing values

signMatrix:

matrix containing the sign of each element in matrix

assayMatrix:

matrix containing additional information about assays, with rows as assays and columns as additional variables

featureMatrix:

matrix containing additional information about features, with rows as features and columns as additional variables

eigenassays:

matrix containing the feature by eigenassay data, with each column in eigenassays corresponding to a left singular vector, representing genome-wide expression, proteome-wide abundance or metabolome-wide intensity in the corresponding eigenassay

eigenexpressions:

numeric vector containing the eigenexpression fraction of each eigenfeature, eigenassay-pair, constituting the diagonal elements of the diagonal matrix connecting the left and right singular values; the diagonal matrix reflects the decoupling and decorrelation of the data, with expression of each eigenfeature restricted to the corresponding eigenassay

eigenfeatures:

matrix containing the eigenfeatures by assay data, with each row corresponding to a right singular vector, representing the expression, abundance or intensity of the corresponding eigenfeature across all assays

assaycorrelations:

matrix containing the correlation between the eigenassays as rows and the assays as columns

featurecorrelations:

matrix containing the correlation between the eigenfeatures as rows and features as columns

fractions:

numeric vector containing the eigenexpression fraction for each eigenfeature, eigenassay-pair, defined as the relative fraction of overall expression that each eigenfeature and eigenassay capture

entropy:

numeric value between 0 and 1 giving the Shannon entropy as measure for data complexity, with an entropy of 0 corresponding to an ordered and redundant data set with all expression captured by a single eigenfeature, eigenassay-pair, and an entropy of 1 corresponding to a disordered and random data set with all eigenfeature, eigenassay-pairs equally expressed

apply:

character containing whether the eigensystem should be computed for the actual data or the variance in the data

excludeEigenfeatures:

numeric vector containing eigenfeature 1 and 2 in case they capture >85% of the data with eigenfeature 2 capturing at least 15%, otherwise numeric value containing eigenfeature 1

colorIdFeatures:

numeric vector or factor containing annotation information on the features

Accessors

matrix(x), matrix(x) <- value

signMatrix(x), signMatrix(x) <- value

assayMatrix(x), assayMatrix(x) <- value

featureMatrix(x), featureMatrix(x) <- value

eigenassays(x), eigenassays(x) <- value

eigenexpressions(x), eigenexpressions(x) <- value

eigenfeatures(x), eigenfeatures(x) <- value

assaycorrelations(x), assaycorrelations(x) <- value

featurecorrelations(x), featurecorrelations(x) <- value

fractions(x), fractions(x) <- value

entropy(x), entropy(x) <- value

apply(x), apply(x) <- value

excludeEigenfeatures(x), excludeEigenfeatures(x) <- value

colorIdFeatures(x), colorIdFeatures(x) <- value

Author(s)

Anneleen Daemen [e-mail], Matthew Brauer [e-mail]

References

Alter O, Brown PO and Botstein D. Singular value decomposition for genome-wide expression data processing and modeling. Proc Natl Acad Sci U.S.A. 97(18), 10101-10106 (2000).

See Also

compute,Eigensystem-method

Examples

## Metabolomics starvation data obtained from http://genomics-pubs.princeton.edu/StarvationMetabolomics/Download.shtml
data(StarvationData)

## An object from class Eigensystem is obtained with the compute method
eigensystem <- compute(StarvationData)

## Obtain entropy
entropy(eigensystem)

biosvd documentation built on April 28, 2020, 6:32 p.m.