casc

casc trains a logisitic regression model on provided single-cell RNAseq clusters. It creates an ROC curve for each cluster vs the others to help decide the appropriate number of clusters.

Installation

casc can be installed via devtools and github:

install.packages("devtools") 
devtools::install_github("jamez-eh/casc")

Useage

casc has the following parameters:

• sce: A SingleCellExperiment with normalized logcounts as an assay or a logcounts matrix with cells as columns and genes as rows.
• clusters: A list of clusterings to evaluate.
• alpha: A parameter for logistic regression. alpha = 1 represents the lasso penalty and alpha = 0 represents the ridge penalty.

registerDoSEQ()
sce_sim <- readRDS("~/sce_sim.rds")
casc_list <- casc(sce = sce_sim, 
                  clusters = list(clustering_1, clustering_2, clustering_3, clustering_4, clustering_5), 
                  alpha = 0.5)

Author

James Hopkins