Fuzzy C-Means (FCM) Clustering

Fuzzy c-means clustering of features by the shape of their z-score trajectories across the "exercise_with_controls" contrasts (the two during contrasts excluded). The pre-computed FCM_CLUSTERS was built with the default arguments.

Usage

run_cmeans(
  DA_list = NULL,
  selected_tissues = c("all", "adipose", "blood", "muscle"),
  selected_omes = c("transcript-rna-seq", "prot-pr", "prot-ol", "prot-ph", "metab"),
  num_clusters_adipose = 13L,
  num_clusters_blood = 12L,
  num_clusters_muscle = 12L,
  modality = c("both", "Endur", "Resist")
)

Arguments

DA_list

list; a named list of data.frame objects, each containing differential analysis results for a specific tissue/assay combination. The list is nested, with tissues at the top level and omes within tissues, or already flattened with names of the form "tissue.assay". If NULL (default), the differential analysis results will be generated with load_differential_analysis. Unless wishing to analyze DA results that are not in MotrpacHumanPreSuspensionAnalysis, this should remain NULL.

selected_tissues

character; passed to load_differential_analysis. One or more of the following: "all", "muscle", "adipose", or "blood".

selected_omes

character; one or more of "transcript-rna-seq", "prot-pr", "prot-ol", "prot-ph", or "metab".

num_clusters_adipose, num_clusters_blood, num_clusters_muscle

integer; the number of clusters for each tissue, a single value each. Defaults to 13 for adipose, 12 for blood, and 12 for muscle, the values used for FCM_CLUSTERS. They were chosen from a sweep of cluster numbers in the motrpac-human-presuspension-repro pipeline (step 13), which plots the minimum centroid distance and related diagnostics for each tissue.

modality

character; which exercise modalities should be used for FCM? One of "both", "Endur", or "Resist".

Value

A named list of objects where names are tissues. Each object is of class "fclust" with additional list component "input" for the matrix of scaled z-scores used as input for FCM (both modalities included, even when modality != "both").

See also

FCM_CLUSTERS, plot_cmeans, run_cluster_cameraPR, run_cluster_ORA

Author

Tyler Sagendorf, Christopher Jin

Examples

if (FALSE) { # \dontrun{
  x1 <- run_cmeans()
  names(x1) # list available components

  # Reuse differential analysis results already in memory
  DA_list <- load_differential_analysis(selected_tissues = "adipose")
  x2 <- run_cmeans(DA_list = DA_list,
                   selected_tissues = "adipose")

  # FCM for a single modality
  x3 <- run_cmeans(selected_tissues = "adipose",
                   modality = "Endur")
} # }