LinearEmbeddingMatrix: LinearEmbeddingMatrix class

View source: R/LEM-other.R

LinearEmbeddingMatrixR Documentation

LinearEmbeddingMatrix class


A description of the LinearEmbeddingMatrix class for storing low-dimensional embeddings from linear dimensionality reduction methods.


LinearEmbeddingMatrix(sampleFactors = matrix(nrow = 0, ncol = 0),
    featureLoadings = matrix(nrow = 0, ncol = 0), factorData = NULL,
    metadata = list())



A matrix-like object of sample embeddings, where rows are samples and columns are factors.


A matrix-like object of feature loadings, where rows are features and columns are factors.


A DataFrame containing factor-level information, with one row per factor.


An optional list of arbitrary content describing the overall experiment.


The LinearEmbeddingMatrix class is a matrix-like object that supports dim, dimnames and as.matrix. It is designed for the storage of results from linear dimensionality reduction methods like principal components analysis (PCA), factor analysis and non-negative matrix factorization.

The sampleFactors slot is intended to store The low-dimensional representation of the samples, such as the principal coordinates from PCA. The feature loadings contributing to each factor are stored in featureLoadings, and should have the same number of columns as sampleFactors. The factorData stores additional factor-level information, such as the percentage of variance explained by each factor, and should have the same number of rows as sampleFactors.

The intended use of this class is to allow PCA and other results to be stored in the reducedDims slot of a SingleCellExperiment object. This means that feature loadings remain attached to the embedding, allowing it to be used in downstream analyses.


A LinearEmbeddingMatrix object is returned from the constructor.


Aaron Lun, Davide Risso and Keegan Korthauer


lem <- LinearEmbeddingMatrix(matrix(rnorm(1000), ncol=5),
    matrix(runif(20000), ncol=5))

LTLA/SingleCellExperiment documentation built on March 24, 2023, 4:07 a.m.