neighborVoting: Runs the neighbor voting algorithm.

In MetaNeighbor: Single cell replicability analysis

Description

The function performs cell type identity prediction based on 'guilt by association' using cross validation. Performance is evaluated by calculating the AUROC for each cell type.

Usage

neighborVoting(
  exp_labels,
  cell_labels,
  network,
  means = TRUE,
  node_degree_normalization = TRUE
)

Arguments

exp_labels	A vector that indicates the dataset source of each sample
cell_labels	sample by cell type matrix that indicates the cell type of each sample (0-absent; 1-present)
network	sample by sample adjacency matrix, ranked and standardized between 0-1
means	default TRUE, determines output formatting
node_degree_normalization	default TRUE, should predictions be divided by node degree?

Value

If means = TRUE (default) a vector containing the mean of AUROC values across cross-validation folds will be returned. If FALSE a list is returned containing a cell type by dataset matrix of AUROC scores, for each fold of cross-validation. Default is over-ridden when more than one cell type is assessed.

See Also

MetaNeighbor

Examples

data("mn_data")
data("GOmouse")
library(SummarizedExperiment)
AUROC_scores = MetaNeighbor(dat = mn_data,
                            experiment_labels = as.numeric(factor(mn_data$study_id)),
                            celltype_labels = metadata(colData(mn_data))[["cell_labels"]],
                            genesets = GOmouse,
                            bplot = TRUE)
AUROC_scores

MetaNeighbor documentation built on Nov. 8, 2020, 5:40 p.m.