ClusterParamSelection: ClusterParamSelection
In nathansam/CircadianTools: Collection of Tools for Detecting Rhythmic Genes

View source: R/cluster_param_selection.R

Calculates validation metrics for different clustering methods and different numbers of partitions. The validation metrics are plotted.

ClusterParamSelection(
  dataset = NULL,
  distance = NULL,
  k = c(2, 5, 10),
  method = c("pam", "agglom", "diana"),
  metric = "euclidean",
  scale = TRUE,
  nthreads = 2,
  save.plot = TRUE,
  save.df = TRUE,
  path = NULL
)

`dataset`	A transcriptomics dataset. Preferably filtered first. First columns should be gene names. All other columns should be expression levels.
`distance`	A distance matrix. If a distance matrix has already been created (such as by using the `DistanceGen` function), the matrix can be passed to this function to save time. If a distance matrix is not provided then it will be generated by the function.
`k`	A numeric vector giving the number of clusters to be evaluated.
`method`	The clustering method(s) to be used. Multiple methods can be considered by providing a character vector. Currently accepts 'pam', 'agglom' and 'diana'.
`metric`	The distance metric to be used to calculate the distances between genes. See parallelDist::parDist for all accepted arguments. Also allows the option of 'abs.correlation'. Not used if a distance matrix is provided.
`scale`	Logical. If TRUE then each gene will be scaled.
`nthreads`	The number of threads to be used for parallel computations. If NULL then the maximum number of threads available will be used.
`save.plot`	Logical. If TRUE then saves the plots generated.
`save.df`	Logical. If TRUE then saves the validation metric results as a .csv file.
`path`	The directory to be used for saving plots and the validation metric results to. Uses the name of the dataset object appended with '_validation' if this argument is not specified.

1
2
3

k.options <- seq(2,10)
hclust.validation <- ClusterParamSelection(Laurasmappings, k=k.options,
                                           nthreads = 2)