Supervised Dimensional Reduction by Guided Partial Least Squares

Description

Guided partial least squares (guided-PLS) is the combination of partial least squares by singular value decomposition (PLS-SVD) and guided principal component analysis (guided-PCA). This package provides implementations of PLS-SVD, guided-PLS, and guided-PCA for supervised dimensionality reduction. The guided-PCA function (new in v1.1.0) automatically handles mixed data types (continuous and categorical) in the supervision matrix and provides detailed contribution analysis for interpretability. For the details of the methods, see the reference section of GitHub README.md <https://github.com/rikenbit/guidedPLS>.

Details

The DESCRIPTION file:

Index of help topics:

PLSSVD                  Partial Least Squares by Singular Value
                        Decomposition (PLS-SVD)
dummyMatrix             Toy model data for using dNMF, dSVD, dsiNMF,
                        djNMF, dPLS, dNTF, and dNTD
guidedPCA               Guided PCA (Principal Component Analysis with
                        Label Guidance)
guidedPLS               Guided Partial Least Squares (guied-PLS)
guidedPLS-package       Supervised Dimensional Reduction by Guided
                        Partial Least Squares
sPLSDA                  Sparse Partial Least Squares Discriminant
                        Analysis (sPLS-DA)
softThr                 Soft-thresholding to make a sparse vector
                        sparse
toyModel                Toy model data for using PLSSVD, sPLSDA, and
                        guidedPLS

Author(s)

NA

Maintainer: NA

References

Le Cao, et al. (2008). A Sparse PLS for Variable Selection when Integrating Omics Data. Statistical Applications in Genetics and Molecular Biology, 7(1)

Reese S E, et al. (2013). A new statistic for identifying batch effects in high-throughput genomic data that uses guided principal component analysis. Bioinformatics, 29(22), 2877-2883

See Also

toyModel,PLSSVD,sPLSDA,guidedPLS

Examples

ls("package:guidedPLS")

Partial Least Squares by Singular Value Decomposition (PLS-SVD)

Description

Two matrices X and Y sharing a row are required

Usage

PLSSVD(X, Y, k=.minDim(X, Y), cortest=FALSE,
  deflation=FALSE, fullrank=TRUE, verbose=FALSE)

Arguments

Value

scoreX : Score matrix which has M-rows and K-columns. loadingX : Loading matrix which has N-rows and K-columns. scoreY : Score matrix which has L-rows and K-columns. loadingY : Loading matrix which has N-rows and K-columns. d : K-length singular value vector of the cross-product matrix X'Y. corX: Correlation Coefficient (Default: NULL) corY: Correlation Coefficient (Default: NULL) pvalX: P-value vector of corX (Default: NULL) pvalY: P-value vector of corY (Default: NULL) qvalX: Q-value vector of BH method against pvalX (Default: NULL) qvalY: Q-value vector of BH method against pvalY (Default: NULL)

Author(s)

Koki Tsuyuzaki

References

Le Cao, et al. (2008). A Sparse PLS for Variable Selection when Integrating Omics Data. Statistical Applications in Genetics and Molecular Biology, 7(1)

Examples

# Test data
data <- toyModel()

# Simple usage
out <- PLSSVD(X=data$X1, Y=data$Y1, k=4)

Toy model data for using dNMF, dSVD, dsiNMF, djNMF, dPLS, dNTF, and dNTD

Description

A label vector is converted to a dummy matrix.

Usage

dummyMatrix(y, center=TRUE)

Arguments

Value

A matrix is generated. The number of row is equal to the length of y and the number of columns is the number of unique elements of y.

Author(s)

Koki Tsuyuzaki

Examples

y <- c(1, 3, 2, 1, 4, 2)
dummyMatrix(y)

Guided PCA (Principal Component Analysis with Label Guidance)

Description

Performs guided PCA by finding principal components that maximize covariance between data matrix X and label/metadata matrix Y. This method extends PLSSVD to automatically handle mixed data types and provide detailed contribution analysis.

Usage

guidedPCA(X, Y, k = NULL, center_X = TRUE, scale_X = TRUE, 
          normalize_Y = TRUE, contribution = TRUE, 
          deflation = FALSE, fullrank = TRUE, verbose = FALSE)

Arguments

Details

The algorithm works as follows:

1. Y preprocessing: Mixed data types in Y are handled automatically: - Categorical variables (factor/character) are converted to dummy variables - Continuous variables (numeric) are used as-is - Logical variables are converted to 0/1 - Missing values are handled (NA in factors become a separate category, NA in numerics become 0)

2. Normalization: When normalize_Y=TRUE (default), each Y column is normalized to unit L2 norm. This ensures equal weight across different metadata types, preventing continuous variables with large scales from dominating categorical ones.

3. Core computation: Computes SVD of the cross-product matrix M = X^T Y, where X is the centered/scaled data matrix and Y is the normalized metadata matrix. This finds linear combinations that maximize covariance between X and Y.

Value

A list of class "guidedPCA" containing:

• loadingX: Loading matrix for X (features x components)

• loadingY: Loading matrix for Y (dummy variables x components)

• scoreX: Score matrix for X (samples x components)

• scoreY: Score matrix for Y (samples x components)

• d: Singular values

• Y_dummy: The dummy-encoded Y matrix used internally

• Y_groups: Group labels for dummy variables

• contrib_features: Feature contributions to each component (if contribution=TRUE)

• contrib_groups: Grouped contributions by original Y variables (if contribution=TRUE)

• variance_explained: Variance explained by each component

Author(s)

Koki Tsuyuzaki

References

Reese S E, et al. A new statistic for identifying batch effects in high-throughput genomic data that uses guided principal component analysis. Bioinformatics, 29(22), 2877-2883, 2013

Examples

# Example with mixed data types
X <- matrix(rnorm(100*50), 100, 50)
Y <- data.frame(
  celltype = factor(sample(c("A", "B", "C"), 100, replace=TRUE)),
  treatment = factor(sample(c("ctrl", "treated"), 100, replace=TRUE)),
  score = rnorm(100)
)
result <- guidedPCA(X, Y, k=3)
print(result)
summary(result)

Guided Partial Least Squares (guied-PLS)

Description

Four matrices X1, X2, Y1, and Y2 are required. X1 and Y1 are supposed to share the rows, X2 and Y2 are supposed to share the rows, and Y1 and Y2 are supposed to share the columns.

Usage

guidedPLS(X1, X2, Y1, Y2, k=.minDim(X1, X2, Y1, Y2),
    cortest=FALSE, fullrank=TRUE, sumcor=FALSE, lambda=1e-6, verbose=FALSE)

Arguments

Value

res: object of svd() loadingYX1: Loading vector to project X1 to lower dimension via Y1 (M times k). loadingYX2: Loading vector to project X2 to lower dimension via Y2 (P times k). scoreX1: Projected X1 (N times k) scoreX2: Projected X2 (O times k) scoreYX1: Projected YX1 (L times k) scoreYX2: Projected YX2 (L times k) corYX1: Correlation Coefficient (Default: NULL) corYX2: Correlation Coefficient (Default: NULL) pvalYX1: P-value vector of corYX1 (Default: NULL) pvalYX2: P-value vector of corYX2 (Default: NULL) qvalYX1: Q-value vector of BH method against pvalYX1 (Default: NULL) qvalYX2: Q-value vector of BH method against pvalYX2 (Default: NULL)

Author(s)

Koki Tsuyuzaki

References

Le Cao, et al. (2008). A Sparse PLS for Variable Selection when Integrating Omics Data. Statistical Applications in Genetics and Molecular Biology, 7(1)

Reese S E, et al. (2013). A new statistic for identifying batch effects in high-throughput genomic data that uses guided principal component analysis. Bioinformatics, 29(22), 2877-2883

Examples

# Test data
data <- toyModel()

# Simple usage
out <- guidedPLS(X1=data$X1, X2=data$X2, Y1=data$Y1, Y2=data$Y2, k=4)

Sparse Partial Least Squares Discriminant Analysis (sPLS-DA)

Description

Two matrices X and Y sharing a row are required

Usage

sPLSDA(X, Y, k=.minDim(X, Y), cortest=FALSE, lambda=1, thr=1e-10, fullrank=TRUE,
    num.iter=10, verbose=FALSE)

Arguments

Value

scoreX : Score matrix which has M-rows and K-columns. loadingX : Loading matrix which has N-rows and K-columns. scoreY : Score matrix which has L-rows and K-columns. loadingY : Loading matrix which has N-rows and K-columns. d : K-length singular value vector of the cross-product matrix X'Y. corX: Correlation Coefficient (Default: NULL) corY: Correlation Coefficient (Default: NULL) pvalX: P-value vector of corX (Default: NULL) pvalY: P-value vector of corY (Default: NULL) qvalX: Q-value vector of BH method against pvalX (Default: NULL) qvalY: Q-value vector of BH method against pvalY (Default: NULL)

Author(s)

Koki Tsuyuzaki

References

Le Cao, et al. (2008). A Sparse PLS for Variable Selection when Integrating Omics Data. Statistical Applications in Genetics and Molecular Biology, 7(1)

Examples

# Test data
data <- toyModel()

# Simple usage
out <- sPLSDA(X=data$X1, Y=data$Y1, k=4)

Soft-thresholding to make a sparse vector sparse

Description

The degree of the sparseness of vector is controlled by the lambda parameter.

Usage

softThr(y, lambda=1)

Arguments

Value

A numerical vector, whose length is the same as that of y.

Author(s)

Koki Tsuyuzaki

Examples

y <- seq(-2, 2, 0.1)
softThr(y)

Toy model data for using PLSSVD, sPLSDA, and guidedPLS

Description

The data is used for confirming the algorithm are properly working.

Usage

toyModel(model="Easy", seeds=123)

Arguments

Value

A list object containing a set of matrices X1, X2, Y1, Y1_dummy, Y2, Y1_dummy.

Author(s)

Koki Tsuyuzaki

See Also

PLSSVD,sPLSDA,guidedPLS

Examples

data <- toyModel(seeds=123)