SickleJr-class {jrSiCKLSNMF}

The SickleJr class

Description

Defines the SickleJr class for use with jrSiCKLSNMF. This object contains all of the information required for analysis using jrSiCKLSNMF. This includes count matrices, normalized matrices, graph Laplacians, hyperparameters, diagnostic plots, and plots of cell clusters.

Value

An object of class SickleJr

Slots

count.matrices

A list containing all of the quality controlled count matrices. Note that these count matrices should not use all features and should only include features that appear in at a minimum 10 cells.

normalized.count.matrices

A list that holds the normalized count matrices

graph.laplacian.list

A list of the graph Laplacians to be used for graph regularization

rowRegularization

A string that indicates the type of row regularization to use. Types include "None" and "L2Norm"

diffFunc

A string that holds the name of the function used to measure the discrepancy between data matrix X and WH for each modality; can be "klp" for the Poisson Kullback-Leibler divergence or "fr" for the Frobenius norm

lambdaWlist

A list of lambda values to use as the hyperparameters for the corresponding \mathbf{W}^v in the v^{\text{th}} modality

lambdaH

A numeric value corresponding to the hyperparameter of the sparsity constraint on \mathbf{H}

Wlist

A list of the generated \mathbf{W}^v matrices, one for each modality

H

The shared \mathbf{H} matrix

WHinitials

A list that if, when using PlotLossvsLatentFactors, all of the cells are used to calculate the initial values, stores these initial generated matrices; can be used as initializations when running RunjrSiCKLSNMF to save time

lossCalcSubsample

A vector that holds the cell indices on which PlotLossvsLatentFactors was calculated

latent.factor.elbow.values

A data frame that holds the relevant information to plot the latent factor elbow plot

minibatch

Indicator variable that states whether the algorithm should use mini-batch updates.

clusterdiagnostics

List of the cluster diagnostic results for the SickleJr object. Includes diagnostic plots from fviz_nbclust and and diagnostics from clValid

clusters

List of results of different clustering methods performed on the SickleJr object

metadata

List of metadata

loss

Vector of the value for the loss function

umap

List of different UMAP-based dimension reductions using umap

plots

Holds various ggplot results for easy access of diagnostics and cluster visualizations