Skip to contents

Pattern Profiler Analysis

Source: R/Pattern_Profiler_Analysis.R
pattern_profiler_analysis.Rd

Complete clustering pipeline starting from a SummarizedExperiment. Writes a parquet file in LONG format for visualisation.

Usage

pattern_profiler_analysis(
  se_proc,
  DEPs_results,
  assay_name = NULL,
  filter_mode = c("any", "all", "specific"),
  alpha = 0.05,
  comparison = NULL,
  condition_order = NULL,
  aggregate = c("median", "mean"),
  c_range = 2:10,
  auto_select_c = TRUE,
  c = NULL,
  selection_method = c("xb", "consensus", "elbow"),
  min_membership = 0.25,
  seed = 42,
  seeds = c(42, 123, 456),
  output_file = NULL,
  verbose = TRUE
)

Arguments

se_proc

SummarizedExperiment with intensity data

DEPs_results

DataFrame with differential expression results (must have an 'Assay' column)

assay_name

Name of the assay to cluster. It is also the value used to filter DEPs_results by its Assay column, so the two must agree. With the default NULL it is taken from DEPs_results$Assay, which records the assay the differential abundance was computed on; supply it explicitly when those results cover more than one assay.

filter_mode

Filtering mode: "any" (significant in at least one comparison), "all" (ALL features, with no significance filtering), "specific" (significant in the comparison given by comparison)

alpha

Significance threshold (default: 0.05)

comparison

Specific comparison (for filter_mode="specific")

condition_order

Condition order for the profiles

aggregate

Aggregation method: "median" or "mean"

c_range

Range of cluster numbers to evaluate

auto_select_c

Automatic selection of the number of clusters (default: TRUE)

c

Fixed number of clusters (if auto_select_c=FALSE)

selection_method

Selection method: "xb", "consensus", "elbow"

min_membership

Minimum membership threshold for inclusion in the output

seed

Integer. Seed for the Mfuzz clustering itself. Fuzzy c-means starts from a random partition, so the cluster labels and the memberships depend on it; fix it to make a run reproducible. The caller's RNG state is restored on exit. Default 42.

seeds

Numeric vector of seeds used when auto_select_c = TRUE: each candidate number of clusters is fitted once per seed and the metrics are averaged, so that the choice of c does not hang on a single partition. Default c(42, 123, 456).

output_file

Path of the output parquet file. NULL by default, which writes nothing to disk; the long-format data is returned regardless in the long_output element of the result.

verbose

Show progress messages

Value

List (invisible) with the clustering results: optimal_c, m, conditions, feature counts, selection_metrics, cluster_counts, the Mfuzz cl object, the standardised eset_std and long_output, the long-format data.frame that is written when output_file is given.

Examples

if (requireNamespace("Mfuzz", quietly = TRUE) &&
    requireNamespace("e1071", quietly = TRUE)) {
  data(nadia_dia)
  res <- process_proteomics(nadia_dia, verbose = FALSE)

  # auto_select_c = TRUE would cluster once per value of c_range
  pp <- pattern_profiler_analysis(
    res$se_proc,
    DEPs_results    = res$DEPs_results,
    assay_name      = "Impseqrob_min",
    filter_mode     = "any",
    condition_order = c("A", "B", "D"),
    auto_select_c   = FALSE,
    c               = 3,
    verbose         = FALSE
  )
  print(pp$cluster_counts)
  print(head(pp$long_output))
}
#>   hard_assignment Freq
#> 1               1  401
#> 2               2   83
#> 3               3  310
#>   FeatureID Cluster Membership        A        B         D
#> 1    P32610       1   0.977237 0.638391 0.514077 -1.152468
#> 2    Q04182       1   0.977214 0.638649 0.513799 -1.152448
#> 3    P21375       1   0.977183 0.638171 0.514313 -1.152484
#> 4    P36090       1   0.977170 0.638774 0.513665 -1.152439
#> 5    P21147       1   0.977150 0.638096 0.514394 -1.152490
#> 6    Q01476       1   0.977112 0.638027 0.514468 -1.152495