robustness: Robustness check for CENA predictions

In mayalevy/CENA: A method for a joint identification of pairwise association together with the particular cell subset of association

Description

Calculate the robustness scores for selected genes.

Usage

robustness(prediction, geneExpressionDataMatrix, phenotypeData, cellSpace,
  genesToRun, no_cores = NULL, numberOfRepeats = 100,
  minCoverage = 0.5)

Arguments

prediction	The result of the original CENA run. The robustness scores are calculated based on the parameters used in this run.
geneExpressionDataMatrix	A matrix containing the single-cell RNA-seq data. Each row corresponds to a certain gene and each column to a certain cell. The algorithm assumes the order of the cells in this scRNA-seq data matrix is the same as the order in the meta-data ('phenotypeData') and the cell-state space ('cellSpace').
phenotypeData	A vector containing the meta-data levels of each of the cells.
cellSpace	Cell coordinates matrix for the single-cell data. Each row represent a cell and each column corresponds to a specific dimension. Only 2-dim spaces are allowed.
genesToRun	A vector of genes for which robustness will be calculated. Due to relatively long running times, it is advised to choose only genes with informative clusters inferred by CENA.
no_cores	The number of cores which will be used for the analysis. The defalt (NULL) is total number of cores minus 1.
numberOfRepeats	Number of CENA runs used to calculate cluster robustness. The default value is 100.
minCoverage	The minimum coverage required for matching between the predicted clusters and the clusters in each iteration. The default value is 0.5.

Value

A matrix where rows are genes and columns are different robustness measures.

• Repetitivity: A fraction representing how much the predicted gene clusters are shared across iterations.

• Normalized ranking mean: Normalized ranking scores mean. Values range between 0 to 1, where a lower value means the perdicted gene cluster is more likely to be selected across iterations.

• Normalized ranking Sd: Normalized ranking scores standard deviation across all robustness iterations.

Examples

data(cellSpace)
data(geneExpressionDataMatrix)
data(phenotypeData)
# running CENA on 5 genes
results = CENA(geneExpressionDataMatrix, phenotypeData, cellSpace, resolution_parameter = 8, no_cores = 1)
genesWithClusters = row.names(geneExpressionDataMatrix)[which(!is.na(rowSums(results$cluster_information)))]
robustnessResults = robustness(results, geneExpressionDataMatrix, phenotypeData, cellSpace, genesToRun = genesWithClusters, no_cores = 1)

mayalevy/CENA documentation built on Jan. 29, 2020, 4:42 p.m.