CEB A Cluster Ensemble Based tool for identifying cell-types

Manual

CEB package

CEB depends on multiple packages, including mclust, fpc, vegan, apcluster, factoextra, ADPclust, SC3, SingleCellExperiment, Rtsne, kernlab. To make sure all the dependency packages are installed.

Download CEB_0.1.1.tar.gz Install R package locally from R studio.

Data Input

Data input provides the interface for users to upload data for the construction of CEB.

RNA data

The expression profiles of single-cell RNA data. The rows of uploaded files should represent RNA, while the columns should represent samples.

Methylation data

The expression profiles of single-cell DNA methylation data. The rows of uploaded files should represent DNA methylation rates (from 0 to 1) summarised at the feature level (i.e. promotes, gene bodies, etc.), while the columns should represent samples.

Just upload the data and parameters to get the result.

CEB_result <- CEB(dataset_list, feature.filter = T, feature.filter.fraction = 0.06, 
                  datatype = "count", log.trans = F,
                  SC3 = T, gene_filter = F, svm_num_cells = 5000, 
                  tSNE = T, dimensions = 3, perplexity = 30, tsne_min_cells = 200, 
                  tsne_min_perplexity = 10, var_genes = NULL,
                  kmeans = T, iter.max = 100,
                  spectral_cluster = T,
                  DBSCAN = F, eps = 270, MinPts = 5,
                  hierarchical_clustering = F, distance_form = "euclidean",
                  SEED = 1)

Reports

Here, we present reports generated by CEB. CEB reports the results of cell-type identifying, and the results can be evaluated by external index if we have the ground truth of cell-types.