Perform clustering analysis from a musica result object

Proportional sample exposures will be used as input to perform clustering.

cluster_exposure(
  result,
  nclust,
  proportional = TRUE,
  method = "kmeans",
  dis.method = "euclidean",
  hc.method = "ward.D",
  clara.samples = 5,
  iter.max = 10,
  tol = 1e-15
)

Arguments

result

A musica_result object generated by a mutational discovery or prediction tool.

nclust

Pre-defined number of clusters.

proportional

Logical, indicating if proportional exposure (default) will be used for clustering.

method

Clustering algorithms. Options are "kmeans" (K-means), "hkmeans" (hybrid of hierarchical K-means), "hclust" (hierarchical clustering), "pam" (PAM), and "clara" (Clara).

dis.method

Methods to calculate dissimilarity matrix. Options are "euclidean" (default), "manhattan", "jaccard", "cosine", and "canberra".

hc.method

Methods to perform hierarchical clustering. Options are "ward.D" (default), "ward.D2", "single", "complete", "average", "mcquitty", "median", and "centroid".

clara.samples

Number of samples to be drawn from dataset. Only used when "clara" is selected. Default is 5.

iter.max

Maximum number of iterations for k-means clustering.

tol

Tolerance level for kmeans clustering level iterations

Value

A one-column data frame with sample IDs as row names and cluster number for each sample.

See also

kmeans

Examples

set.seed(123)
data(res_annot)
clust_out <- cluster_exposure(res_annot, nclust = 2)
#> Metric: 'euclidean'; comparing: 7 vectors.
#> Warning: FANNY algorithm has not converged in 'maxit' = 10 iterations