report_cluster_metrics: Reports BEARscc metrics for clusters.

Description Usage Arguments Value Author(s) Examples

Description

To quantitatively evaluate the results, three metrics are calculated from the noise consensus matrix: 'stability' is the average frequency with which cells within a cluster associate with each other across simulated replicates; 'promiscuity' measures the association frequency between cells within a cluster and those outside of it; and 'score' is the difference between 'stability' and 'promiscuity'. Importantly, 'score' reflects the overall "robustness" of a cluster to technical variance. These metrics may be calculated on cell or cluster-wise basis; here, they are calculated cluster-wise.

Usage

1
2
report_cluster_metrics(cluster_labels, consensus_matrix,
    weighted_mean = FALSE, plot = FALSE, file = "Rplot")

Arguments

cluster_labels

Cluster labels for each cell across various cluster numbers and the original clustering.

consensus_matrix

A noise consensus output by compute_consensus()

weighted_mean

A flag indicating whether to weigh observed clusters evenly or scale them by the number of samples in the cluster.

plot

A flag to determine whether to plot the boxplot of cluster metrics evaluated from the noise consensus with root file.

file

A string indicating the root desired for the resulting plots of the function.

Value

A melted data.frame of BEARscc metrics for each cluster:

[,1] "Cluster.identity" The number of the cluster within the respective clustering.
[,2] "Cluster.size" Number of samples in the cluster.
[,3] "Metric" Whether the metric is the BEARscc score, promiscuity, or stability.
[,4] "Value" Value of the relevant BEARscc metric for the cluster in a clustering.
[,5] "Clustering" The clustering pertinant to the cell-wise metrics described.
[,6] "Singlet" A binary output concerning whether the cluster consists of a single sample.
[,7] "Clustering.Mean" The average of the respective metric across cells of the clsuter.

Author(s)

David T. Severson <david_severson@hms.harvard.edu>

Maintainer: Benjamin Schuster-Boeckler <benjamin.schuster-boeckler@ludwig.ox.ac.uk>

Examples

1
2
3
4
5
data(analysis_examples)

cluster_scores.df <- report_cluster_metrics(BEARscc_clusts.df, noise_consensus,
    plot=TRUE, file="example")
cluster_scores.df

seversond12/BEARscc documentation built on June 8, 2020, 8:04 a.m.