immunarch (version 0.10.3)

immunr_hclust: Clustering of objects or distance matrices

Description

Clusters the data with one of the following methods:

• immunr_hclust clusters the data using the hierarchical clustering from hcut;

• immunr_kmeans clusters the data using the K-means algorithm from kmeans;

• immunr_dbscan clusters the data using the DBSCAN algorithm from dbscan.

Usage

immunr_hclust(.data, .k = 2, .k.max = nrow(.data) - 1, .method = "complete", .dist = TRUE)
immunr_kmeans(.data, .k = 2, .k.max = as.integer(sqrt(nrow(.data))) + 1,
.method = c("silhouette", "gap_stat"))
immunr_dbscan(.data, .eps, .dist = TRUE)

Value

immunr_hclust - list with two elements. The first element is an output from factoextra::hcut. The second element is an output from factoextra::fviz_nbclust

immunr_kmeans - list with three elements. The first element is an output from kmeans. The second element is an output from factoextra::fviz_nbclust. The third element is the input dataset .data.

immunr_dbscan - list with two elements. The first element is an output from fpc::dbscan. The second element is the input dataset .data.

Arguments

.data

Matrix or data frame with features, distance matrix or output from repOverlapAnalysis or geneUsageAnalysis functions.

.k

The number of clusters to create, defined as k to hcut or as centers to kmeans.

.k.max

Limits the maximum number of clusters. It is passed as k.max to factoextra::fviz_nbclust for immunr_hclust and immunr_kmeans.

.method

Passed to factoextra::hcut or as factoextra::fviz_nbclust.

In case of factoextra::hcut the agglomeration method is going to be used (argument hc_method).

In case of factoextra::fviz_nbclust it is the method to be used for estimating the optimal number of clusters (argument method).

.dist

If TRUE then ".data" is expected to be a distance matrix. If FALSE then the euclidean distance is computed for the input objects.

.eps

Local radius for expanding clusters, minimal distance between points to expand clusters. Passed as eps to dbscan.

Examples

data(immdata)
gu <- geneUsage(immdata$data, .norm = TRUE)
immunr_hclust(t(as.matrix(gu[, -1])), .dist = FALSE)

gu[is.na(gu)] <- 0
immunr_kmeans(t(as.matrix(gu[, -1])))