cellCellRanks: Rank estimation of the CCI-tensor

Description Usage Arguments Value Author(s) See Also Examples

Description

SVD is performed in each mode.

Usage

1
2
3
4
cellCellRanks(sce, centering=TRUE,
    mergeas=c("mean", "sum"), outerfunc=c("*", "+"), comb=c("random", "all"),
    num.sampling=100, num.perm=1000, assayNames = "counts", verbose=FALSE,
    num.iter1=5, num.iter2=5, num.iter3=NULL)

Arguments

sce

A object generated by instantization of SingleCellExperiment-class.

centering

When the value is TRUE, input matrix is summarized as celltype-level vectors (Default: TRUE).

mergeas

When the centering is TRUE, "mean" (celltype-level mean vector) or "sum" (celltype-level sum vector) is calculated (Default: "mean").

outerfunc

When the centering is TRUE, "*" (Kronecker product) or "+" (Kronecker sum) or is calculated (Default: "+").

comb

When the centering is FALSE, "random" (random cell-cell pairing) or "all" (all possible cell-cell pairing) is calculed (Default: "random").

num.sampling

The number of random sampling used (Default: 100).

num.perm

The number of the permutation in label permutation test (Default: 1000).

assayNames

The unit of gene expression for using scTensor (e.g. normcounts, cpm...etc) (Default: "counts").

verbose

The verbose parameter for nnTensor::NTD (Default: FALSE).

num.iter1

The number of iteration to estimate the rank of mode-1 matricised data tensor (Default: 5).

num.iter2

The number of iteration to estimate the rank of mode-2 matricised data tensor (Default: 5).

num.iter3

The number of iteration to estimate the rank of mode-3 matricised data tensor (Default: NULL).

Value

RSS: A list with three elements, in which each element means the average reconstructed error in each rank. selected: A vector with three elements, in which each element means the estimated ranks in mode-1, 2 and 3 matricization.

Author(s)

Koki Tsuyuzaki

See Also

SingleCellExperiment.

Examples

1
showMethods("cellCellRanks")

rikenbit/scTensor documentation built on Dec. 16, 2021, 5:51 p.m.