Ridge Redundancy Analysis for High-Dimensional Omics Data
Generate a list of rank-specific Bhat matrices (the coefficient of Rid...
Compute the components of the coefficient Bhat using SVD.
Estimate an appropriate value for the ridge penalty (lambda).
Generate rank-specific matrices by combining the left and right compon...
Compute MSE for different ranks of the coefficient Bhat and lambda.
Generate simulated data for Ridge Redundancy Analysis (RDA).
Generate simulated data for Ridge Redundancy Analysis (RDA).
Calculate the Bhat matrix from the return of the rrda.fit function.
Cross-validation for Ridge Redundancy Analysis
Calculate the coefficient Bhat by Ridge Redundancy Analysis.
Heatmap of the results of cross-validation for Bhat obtained from the ...
Plot the results of cross-validation for Bhat obtained from the `rrda....
Calculate the predicted matrix Yhat using the coefficient Bhat obtaine...
Summarize the results of cross-validation for the coefficient Bhat obt...
Top feature interactions visualization with rank and lambda penalty
Compute the square root of the inverse of (d^2 + lambda).
Scale a matrix using unbiased estimators for the mean and standard dev...
Unscale a matrix based on provided mean and standard deviation values.
Apply unscaling to a nested list of matrices using specified mean and ...
Generate a list of rank-specific Yhat matrices.
Efficient framework for ridge redundancy analysis (rrda), tailored for high-dimensional omics datasets where the number of predictors exceeds the number of samples. The method leverages Singular Value Decomposition (SVD) to avoid direct inversion of the covariance matrix, enhancing scalability and performance. It also introduces a memory-efficient storage strategy for coefficient matrices, enabling practical use in large-scale applications. The package supports cross-validation for selecting regularization parameters and reduced-rank dimensions, making it a robust and flexible tool for multivariate analysis in omics research. Please refer to our article (Yoshioka et al., 2025) for more details.