Run Joint-RPCA and store embedding in reducedDim
Universal Joint RPCA Wrapper
Optional name of an alternative experiment. If supplied,
Joint-RPCA is run on altExp(x, altexp) instead of x.
Character scalar giving the name of the reducedDim slot
in which to store the joint sample embedding. Defaults to "JointRPCA".
Input object: MultiAssayExperiment, SummarizedExperiment
(including TreeSummarizedExperiment), list of matrices, or single matrix.
Character vector of experiment names to extract when x
is a MultiAssayExperiment (i.e. names(experiments(x))).
If NULL, all experiments are used.
For MultiAssayExperiment inputs, one assay per experiment is
used: by default the first assay returned by
assayNames() (or index 1L if unnamed).
The actually used assay names are recorded in $assay_names_used in
the result. If you need a different assay (e.g. "relab" instead of
"counts"), subset or reorder assays in x before calling
jointRPCAuniversal().
Character string specifying preprocessing applied to each
input table before ordination. Use "rclr" to apply the robust CLR
transform (via decostand(method = "rclr")) or "none" to
disable transformation (data are used as-is after masking non-finite values).
Numeric tolerance passed to optspace().
Logical; whether to center the reconstructed low-rank matrix (double-centering) prior to SVD/PCA steps.
Logical; whether to scale the reconstructed matrix prior to
SVD/PCA steps. Defaults to FALSE.
Additional arguments passed to .joint_rpca().
The input object x with a new entry in
reducedDim(x, name) containing the Joint-RPCA sample embedding.
The full Joint-RPCA result (including distances, cross-validation
statistics and transformed tables) is stored in
metadata(x)$JointRPCA[[name]].
Output from .joint_rpca() with extra fields when x is a
MultiAssayExperiment:
$experiment_names: character vector of experiments used.
$assay_names_used: named character vector giving, for each
experiment, the assay name that was used (typically the first in
assayNames()).