Package 'hcocena'

Description

Creates an annotation column that combines donor meta-cluster assignment and timepoint labels (e.g. MC1__24h). This column can be used as grouping_v in hc_run_expression_analysis_2() to build a cluster heatmap by meta-cluster x timepoint.

Usage

hc_add_meta_time_grouping(
  hc,
  donor_col = "Subject",
  time_col = "Time_token",
  grouping_col = "meta_cluster_time",
  slot_name = "longitudinal_endotypes",
  layer = "all",
  time_levels = NULL,
  meta_cluster_levels = NULL,
  separator = "__",
  append_layer_suffix = FALSE
)

Arguments

Value

Updated HCoCenaExperiment. The computed order is stored in hc@satellite[[slot_name]]$meta_time_grouping and/or the resolved per-layer step 2 slots.

Examples

hc <- hc_example_data("clustered")
hc <- hc_longitudinal_step1_module_donor(
  hc,
  donor_col = "donor",
  time_col = "timepoint",
  time_levels = c("T1", "T2"),
  k = 2,
  nstart = 1,
  cap_runs = 1,
  impute = FALSE,
  seed = 1
)$hc
hc <- hc_longitudinal_endotype_clustering(hc, k = 2, nstart = 1, seed = 1)
hc <- hc_longitudinal_meta_clustering(
  hc,
  feature_source = "feature_matrix_used",
  k = 2:3,
  method = "kmeans",
  consensus = TRUE,
  seed = 1
)
hc <- hc_add_meta_time_grouping(hc, donor_col = "donor", time_col = "timepoint")

Description

Alluvial comparison plots (S4 API)

Usage

hc_algo_alluvial(hc)

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("clustered")
hc <- hc_algo_alluvial(hc)

Description

Inverse of base::log() for an arbitrary base, i.e. hc_antilog(x, 2) returns 2^x.

Usage

hc_antilog(lx, base)

Arguments

Value

The de-logged values, in the same shape as lx.

Examples

hc_antilog(c(1, 2, 3), base = 2)

Description

Resolve and apply standard paths in one call

Usage

hc_auto_set_paths(
  hc,
  project_root_hint = NULL,
  local_count_subdir = "count_data",
  local_annotation_subdir = "annotation_data",
  local_reference_subdir = "reference_files",
  local_output_subdir = "output",
  docker_reference_dir = "/home/rstudio/reference_files",
  docker_count_dir = "/home/rstudio/count_data",
  docker_annotation_dir = "/home/rstudio/annotation_data",
  docker_output_dir = "/home/rstudio/output",
  create_docker_dirs = TRUE,
  fallback_count_data = "PATH_TO_FOLDER_THAT_HOLDS_YOUR_GENE_EXPRESSION_TABLES/",
  fallback_annotation = "PATH_TO_FOLDER_THAT_HOLDS_YOUR_ANNOTATION_TABLES/",
  fallback_reference_files = "PATH_TO_REFERENCE_FILES_FOLDER/",
  fallback_output = "PATH_TO_FOLDER_WHERE_HCOCENA_SHOULD_SAVE_ALL_ANALYSIS_OUTPUTS/",
  dir_count_data = NULL,
  dir_annotation = NULL,
  dir_reference_files = NULL,
  dir_output = NULL
)

Arguments

Value

Updated HCoCenaExperiment.

Examples

root <- file.path(tempdir(), "hcocena-auto-paths")
dir.create(file.path(root, "reference_files"), recursive = TRUE, showWarnings = FALSE)
dir.create(file.path(root, "output"), recursive = TRUE, showWarnings = FALSE)
hc <- hc_init()
hc <- hc_auto_set_paths(
  hc,
  project_root_hint = root,
  local_count_subdir = NULL,
  local_annotation_subdir = NULL
)
as.data.frame(methods::slot(hc_config(hc), "paths"))

Description

hc_auto_tune() recommends and optionally applies:

1. Layer-wise cutoffs based on run_expression_analysis_1() output.

2. Clustering settings based on modularity across candidate algorithms.

Usage

hc_auto_tune(
  hc,
  tune_cutoff = TRUE,
  tune_clustering = TRUE,
  apply = TRUE,
  cluster_algorithms = c("auto", "cluster_leiden", "cluster_louvain"),
  resolution_grid = base::seq(0.05, 0.5, by = 0.05),
  no_of_iterations = 2,
  partition_type = "RBConfigurationVertexPartition",
  max_cluster_count_per_gene = 1,
  module_count_bounds = c(3, 60),
  prefer_resolution = 0.1,
  verbose = TRUE
)

Arguments

Details

The tuning report is stored in hc@satellite$auto_tune.

Value

Updated HCoCenaExperiment with tuning report in hc@satellite$auto_tune.

Examples

hc <- hc_example_data("after_part1")
hc <- hc_auto_tune(hc, apply = FALSE)

Description

Alternative cutoff tuning method that can be compared against the current simple cutoff recommendation.

Usage

hc_auto_tune_cutoff_tiered(
  hc,
  apply = FALSE,
  tier1_sft_min = 0.85,
  tier1_mean_k_min = 10,
  tier2_sft_min = 0.8,
  tier2_mean_k_min = 5,
  fallback_sft_min = 0.7,
  require_negative_slope = TRUE,
  max_no_networks = NULL,
  max_components = NULL,
  min_node_fraction = NULL,
  fallback_respect_base_filters = TRUE,
  verbose = TRUE
)

Arguments

Details

For each layer, the function:

1. Summarizes cutoff statistics.

2. Computes additional diagnostics (slope, mean_k).

3. Applies a staged rule set (strict, relaxed, best_available).

Results are stored in hc@satellite$cutoff_tuning (and mirrored to hc@satellite$auto_tune_cutoff_tiered for compatibility).

Value

Updated HCoCenaExperiment with a report in hc@satellite$cutoff_tuning.

Examples

hc <- hc_example_data("after_part1")
hc <- hc_auto_tune_cutoff_tiered(hc, apply = FALSE, verbose = FALSE)

Description

The previously constructed layer-specific networks are being integrated.

Usage

hc_build_integrated_network(
  hc,
  mode = "u",
  with = NULL,
  multi_edges = "min",
  GFC_when_missing = NULL
)

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("after_part2")
hc <- hc_build_integrated_network(hc, mode = "u")

Description

Module cell-type activity from Enrichr markers via decoupleR (S4 API)

Usage

hc_celltype_activity_decoupler(
  hc,
  databases = c("Descartes_Cell_Types_and_Tissue_2021", "Human_Gene_Atlas"),
  custom_gmt_files = NULL,
  clusters = c("all"),
  mode = c("coarse", "fine"),
  top = 3,
  qval = 0.1,
  padj = "BH",
  activity_input = c("gfc", "fc", "expression"),
  fc_comparisons = NULL,
  method = "ulm",
  minsize = 5,
  min_term_genes = 5,
  annotation_slot = c("auto", "enriched_per_cluster", "enriched_per_cluster2"),
  slot_suffix = "decoupler",
  clear_previous_slots = FALSE,
  coarse_map = NULL,
  coarse_include_other = TRUE,
  refresh_db = FALSE,
  export_excel = TRUE,
  excel_file = "Module_Celltype_Activity_decoupler.xlsx",
  plot_heatmap = FALSE,
  heatmap_file_name = "Heatmap_modules_celltype_activity_decoupler.pdf",
  ...
)

Arguments

Value

Updated HCoCenaExperiment.

Examples

if (requireNamespace("decoupleR", quietly = TRUE)) {
  hc <- hc_example_data("clustered")
  gmt <- system.file("extdata", "toy_celltype_markers.gmt",
                     package = "hcocena")
  hc <- hc_celltype_activity_decoupler(
    hc,
    databases = character(0),
    custom_gmt_files = c(CellTypes = gmt),
    export_excel = FALSE
  )
  names(hc_satellite(hc, "celltype_activity_decoupler"))
}

Description

Module cell-type annotation from Enrichr (S4 API)

Usage

hc_celltype_annotation(
  hc,
  databases = c("Descartes_Cell_Types_and_Tissue_2021", "Human_Gene_Atlas"),
  custom_gmt_files = NULL,
  clusters = c("all"),
  mode = c("coarse", "fine"),
  top = 3,
  qval = 0.1,
  padj = "BH",
  min_term_genes = 5,
  min_gs_size = 10,
  max_gs_size = 5000,
  annotation_slot = c("auto", "enriched_per_cluster", "enriched_per_cluster2"),
  slot_suffix = NULL,
  clear_previous_slots = TRUE,
  coarse_map = NULL,
  coarse_include_other = TRUE,
  refresh_db = FALSE,
  export_excel = TRUE,
  excel_file = "Module_Celltype_Annotation.xlsx",
  plot_heatmap = FALSE,
  heatmap_file_name = "Heatmap_modules_celltype_annotation.pdf",
  ...
)

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("clustered")
gmt <- system.file("extdata", "toy_celltype_markers.gmt", package = "hcocena")
# A local marker GMT instead of the default Enrichr libraries, so that the
# example runs offline:
hc <- hc_celltype_annotation(
  hc,
  databases = character(0),
  custom_gmt_files = c(CellTypes = gmt),
  export_excel = FALSE
)
hc_satellite(hc, "celltype_annotation")$selected_celltypes

Description

Recalculate grouped GFCs and replot module heatmap (S4 API)

Usage

hc_change_grouping_parameter(
  hc,
  group_by,
  col_order = NULL,
  cluster_columns = FALSE,
  row_order = NULL,
  cluster_rows = TRUE
)

Arguments

Details

Uses the existing module definitions from the current object and only recalculates grouped GFCs for a different grouping variable. No integrated network rebuild and no module re-clustering is performed.

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("clustered")
hc <- hc_change_grouping_parameter(hc, group_by = "batch")

Description

Iteratively calls fix_dir() on the provided working directory paths to fix them if necessary or throw an error if they are invalid.

Usage

hc_check_dirs(hc, create_output_dir = TRUE)

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("prepared")
hc <- hc_set_paths(
  hc,
  dir_count_data = FALSE,
  dir_annotation = FALSE,
  dir_reference_files = FALSE,
  dir_output = tempdir()
)
hc <- hc_check_dirs(hc)

Description

Check TF targets (S4 API)

Usage

hc_check_tf(hc, TF)

Arguments

Value

Updated HCoCenaExperiment.

Examples

# Needs the ChEA3 results of hc_tf_overrep_network(), which queries the
# ChEA3 web service.
hc <- hc_example_data("clustered")
hc <- hc_tf_overrep_network(hc, topTF = 10, topTarget = 5)
tfs <- names(hc_satellite(hc, "tf_network_targets"))
if (length(tfs) > 0) {
  hc <- hc_check_tf(hc, TF = tfs[[1]])
}

Description

The function offers several community detection algorithms to identify dense regions in the co-expression network. These dense regions represent collections of highly co-expressed genes likely to form a functional group.

Usage

hc_cluster_calculation(
  hc,
  cluster_algo = "cluster_leiden",
  no_of_iterations = 2,
  resolution = 0.1,
  partition_type = "RBConfigurationVertexPartition",
  max_cluster_count_per_gene = 1,
  return_result = FALSE
)

Arguments

Value

Updated HCoCenaExperiment. If return_result = TRUE and the clustering backend returns a cluster table, that table is returned instead.

Examples

hc <- hc_example_data("clustered")
hc <- hc_cluster_calculation(hc, cluster_algo = "cluster_leiden")

Description

Access cluster information from an HCoCenaExperiment

Usage

hc_clusters(x)

## S4 method for signature 'HCoCenaExperiment'
hc_clusters(x)

Arguments

Value

Cluster information or NULL if missing.

Examples

hc <- hc_init()
hc_clusters(hc)

Description

Add categorical module heatmap annotations (S4 API)

Usage

hc_col_anno_categorical(hc, variables, variable_label = NULL, type = "abs")

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("clustered")
hc <- hc_col_anno_categorical(hc, variables = "batch")

Description

Adds numeric annotation tracks to the columns of the module heatmap. Call with variables = NULL to remove previously added tracks.

Usage

hc_col_anno_numerical(hc, variables, variable_label)

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("clustered")
hc <- hc_col_anno_numerical(hc, variables = "batch", variable_label = "batch")

Description

Re-plots the integrated network with one module emphasised by colour and node size.

Usage

hc_colour_single_cluster(hc, cluster)

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("clustered")
hc <- hc_colour_single_cluster(hc, cluster = "gold")

Description

Access configuration from an HCoCenaExperiment

Usage

hc_config(x)

## S4 method for signature 'HCoCenaExperiment'
hc_config(x)

Arguments

Value

A HCoCenaConfig object.

Examples

hc <- hc_init()
class(hc_config(hc))

Description

Assigns new sample group labels from the data structure rather than from metadata, by cutting a hierarchical clustering tree. Useful when the variable of interest does not match the structure seen in the heatmaps or the PCA.

Usage

hc_cut_hclust(
  hc,
  by = "all",
  set = "all",
  method = "complete",
  k,
  save = FALSE
)

Arguments

Details

Re-run with save = FALSE to try different k values; only set save = TRUE once you have settled on one, since that overwrites the original groups and exports a new annotation file (without row names). After saving, re-run the pipeline from hc_run_expression_analysis_2() onward.

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("clustered")
hc <- hc_cut_hclust(hc, by = "module", k = c(2, 2), save = FALSE)

Description

Define layers in an HCoCenaExperiment

Usage

hc_define_layers(hc, data_sets = list())

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_init()
hc <- hc_define_layers(
  hc,
  data_sets = list(
    Layer1 = c("counts.tsv", "anno.tsv")
  )
)
as.data.frame(methods::slot(hc_config(hc), "layer"))

Description

Returns one of the small example objects shipped in inst/extdata, taken at a given stage of the standard workflow, so that individual steps can be tried without running everything before them. The data are two simulated layers of 40 marker genes for four blood cell types over 32 samples; see inst/scripts/make-fixtures.R for the generating code.

Usage

hc_example_data(
  stage = c("prepared", "after_part1", "after_part2", "clustered"),
  dir_output = NULL
)

Arguments

Details

The output directory stored in the object is replaced by a fresh directory inside tempdir(), so functions that write plots or tables never write outside the session's temporary directory.

Value

A HCoCenaExperiment.

Examples

hc <- hc_example_data("clustered")
hc
table(hc_gene_to_cluster(hc)$color)

Description

Export clusters (S4 API)

Usage

hc_export_clusters(hc)

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("clustered")
hc <- hc_export_clusters(hc)

Description

Export network to Cytoscape (S4 API)

Usage

hc_export_to_cytoscape(hc, name = "my igraph", docker_container = FALSE)

Arguments

Value

Updated HCoCenaExperiment.

Examples

# Needs a running Cytoscape instance reachable through RCy3.
cytoscape_up <- requireNamespace("RCy3", quietly = TRUE) &&
  isTRUE(tryCatch({
    RCy3::cytoscapePing()
    TRUE
  }, error = function(e) FALSE))
if (cytoscape_up) {
  hc <- hc_example_data("clustered")
  hc <- hc_export_to_cytoscape(hc, name = "hcocena example")
}

Description

Export network to local folder (S4 API)

Usage

hc_export_to_local_folder(hc, file)

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("clustered")
hc <- hc_export_to_local_folder(hc, file = tempdir())

Description

Determines hub genes per module from a combined ranking of weighted degree, closeness and betweenness centrality.

Usage

hc_find_hubs(
  hc,
  clusters = c("all"),
  top = 10,
  tree_layout = FALSE,
  TF_only = FALSE,
  save = FALSE,
  plot = FALSE
)

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("clustered")
hc <- hc_find_hubs(hc)

Description

Functional enrichment (S4 API)

Usage

hc_functional_enrichment(
  hc,
  gene_sets = c("Go", "Kegg", "Hallmark", "Reactome"),
  custom_gmt_files = NULL,
  top = 5,
  clusters = c("all"),
  padj = "BH",
  qval = 0.05,
  universe = c("all_genes", "network"),
  consistent_terms = TRUE,
  heatmap_side = "left",
  heatmap_cluster_rows = FALSE,
  cluster_columns = FALSE,
  heatmap_cluster_columns = NULL,
  heatmap_show_row_dend = FALSE,
  heatmap_show_column_dend = FALSE,
  col_order = NULL,
  heatmap_col_order = NULL,
  heatmap_order = NULL,
  heatmap_module_label_mode = "same",
  heatmap_show_gene_counts = FALSE,
  heatmap_column_label_fontsize = NULL,
  heatmap_module_label_fontsize = NULL,
  legend_fontsize = NULL,
  enrichment_label_fontsize = NULL,
  enrichment_db_header_fontsize = NULL,
  enrichment_label_wrap = FALSE,
  enrichment_label_wrap_width = 30,
  gfc_scale_limits = NULL,
  store_panel_objects = c("auto", "always", "never"),
  pdf_width = NULL,
  pdf_height = NULL,
  pdf_pointsize = 11,
  ...
)

Arguments

Value

Updated HCoCenaExperiment.

Significance columns

The result tables carry pvalue (raw), p_adjusted (corrected with the method given in padj, and what the qval filter and the ranking use), padj_method (which correction that was) and q_storey (clusterProfiler's Storey q-value, NA where pi0 could not be fitted - which is common for small gene-set collections). qvalue is kept as an alias of p_adjusted so existing scripts and exports keep working.

Examples

hc <- hc_example_data("clustered")
gmt <- tempfile(fileext = ".gmt")
writeLines(
  c(
    paste(c("T cells", "-", "CD3D", "CD3E", "CD3G", "CD2", "CD28",
            "LCK", "ZAP70", "IL7R", "CD7", "TRAC"), collapse = "\t"),
    paste(c("B cells", "-", "CD19", "MS4A1", "CD79A", "CD79B", "BLNK",
            "PAX5", "CR2", "FCRL1", "TNFRSF13B", "VPREB3"), collapse = "\t"),
    paste(c("Monocytes", "-", "CD14", "LYZ", "FCN1", "VCAN", "S100A8",
            "S100A9", "CSF1R", "ITGAM", "CD68", "FCGR3A"), collapse = "\t"),
    paste(c("NK cells", "-", "NKG7", "GNLY", "KLRD1", "KLRF1", "PRF1",
            "GZMB", "NCR1", "KLRC1", "FGFBP2", "SPON2"), collapse = "\t")
  ),
  gmt
)
hc <- hc_functional_enrichment(
  hc,
  gene_sets = character(0),
  custom_gmt_files = c(CellTypes = gmt)
)

Description

Returns the mapping of every network gene to the module it was assigned to.

Usage

hc_gene_to_cluster(hc)

Arguments

Value

A data frame with two columns: gene (gene symbol) and color (module colour). Genes that were not assigned to any module carry the colour "white".

Examples

hc <- hc_example_data("clustered")
head(hc_gene_to_cluster(hc))

Description

Returns the column-order vector stored by hc_add_meta_time_grouping().

Usage

hc_get_meta_time_col_order(
  hc,
  slot_name = "longitudinal_endotypes",
  layer = NULL
)

Arguments

Value

Character vector with heatmap column order as expected by hc_change_grouping_parameter().

Examples

hc <- hc_example_data("clustered")
hc <- hc_longitudinal_step1_module_donor(
  hc,
  donor_col = "donor",
  time_col = "timepoint",
  time_levels = c("T1", "T2"),
  k = 2,
  nstart = 1,
  cap_runs = 1,
  impute = FALSE,
  seed = 1
)$hc
hc <- hc_longitudinal_endotype_clustering(hc, k = 2, nstart = 1, seed = 1)
hc <- hc_longitudinal_meta_clustering(
  hc,
  feature_source = "feature_matrix_used",
  k = 2:3,
  method = "kmeans",
  consensus = TRUE,
  seed = 1
)
hc <- hc_add_meta_time_grouping(hc, donor_col = "donor", time_col = "timepoint")
hc_get_meta_time_col_order(hc)

Description

Module scores (S4 API)

Usage

hc_get_module_scores(hc, save = TRUE)

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("clustered")
hc <- hc_get_module_scores(hc)

Description

Access the integrated network of an HCoCenaExperiment

Usage

hc_graph(x)

## S4 method for signature 'HCoCenaExperiment'
hc_graph(x)

Arguments

Value

The integrated network as an igraph object, or NULL if hc_build_integrated_network() has not been run yet.

Examples

hc <- hc_example_data("clustered")
g <- hc_graph(hc)
igraph::vcount(g)

Description

Re-plots the integrated network with a gene set of interest highlighted.

Usage

hc_highlight_geneset(
  hc,
  gene_set,
  name = NULL,
  col = "black",
  label_offset = 3,
  plot = TRUE,
  save = TRUE
)

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("clustered")
hc <- hc_highlight_geneset(hc, gene_set = c("CD3D", "CD19"))

Description

Replaces the current module assignment with one read from a file, e.g. a clustering exported earlier with hc_export_clusters().

Usage

hc_import_clusters(hc, file, sep = "\t", header = TRUE)

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("clustered")
gtc <- hc_gene_to_cluster(hc)
f <- tempfile(fileext = ".tsv")
utils::write.table(gtc, f, sep = "\t", row.names = FALSE, quote = FALSE)
hc <- hc_import_clusters(hc, file = f)

Description

Import layout from Cytoscape (S4 API)

Usage

hc_import_layout_from_cytoscape(hc)

Arguments

Value

Updated HCoCenaExperiment.

Examples

# Needs a running Cytoscape instance reachable through RCy3.
cytoscape_up <- requireNamespace("RCy3", quietly = TRUE) &&
  isTRUE(tryCatch({
    RCy3::cytoscapePing()
    TRUE
  }, error = function(e) FALSE))
if (cytoscape_up) {
  hc <- hc_example_data("clustered")
  hc <- hc_export_to_cytoscape(hc, name = "hcocena example")
  # ... arrange the network in Cytoscape, then:
  hc <- hc_import_layout_from_cytoscape(hc)
}

Description

Import Cytoscape layout from local folder (S4 API)

Usage

hc_import_layout_from_local_folder(hc, file)

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("clustered")
# A layout CSV as written by hc_import_layout_from_cytoscape(): one row per
# node, named by gene, with x and y coordinates.
g <- hc_graph(hc)
xy <- igraph::layout_with_fr(g)
rownames(xy) <- igraph::V(g)$name
layout_file <- file.path(tempdir(), "network_layout.csv")
utils::write.csv(xy, layout_file)
hc <- hc_import_layout_from_local_folder(hc, file = layout_file)

Description

Initialize an empty HCoCenaExperiment

Usage

hc_init()

Value

A valid, empty HCoCenaExperiment.

Examples

hc <- hc_init()
methods::is(hc, "HCoCenaExperiment")

Description

A folder with the given name is created in the output directory. All analysis outputs will be saved to this folder.

Usage

hc_init_save_folder(hc, name, use_output_dir = FALSE)

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("prepared")
hc <- hc_init_save_folder(hc, name = "")

Description

Access integration payload from an HCoCenaExperiment

Usage

hc_integration(x)

## S4 method for signature 'HCoCenaExperiment'
hc_integration(x)

Arguments

Value

A HCoCenaIntegration object.

Examples

hc <- hc_init()
class(hc_integration(hc))

Description

Access layer results from an HCoCenaExperiment

Usage

hc_layer_results(x)

## S4 method for signature 'HCoCenaExperiment'
hc_layer_results(x)

Arguments

Value

A S4Vectors::SimpleList.

Examples

hc <- hc_init()
class(hc_layer_results(hc))

Description

Lists Enrichr libraries and optionally filters for cell/tissue related names.

Usage

hc_list_celltype_databases(
  pattern = NULL,
  include_all = FALSE,
  ignore_case = TRUE
)

Arguments

Value

A data frame with Enrichr library metadata.

Examples

hc_list_celltype_databases()

Description

Queries the model-listing endpoint of each selected provider and returns the available models as a data frame. Supports the same providers as hc_llm_enrichment(): Gemini, Claude (Anthropic), OpenAI/ChatGPT, and a local OpenAI-compatible server (vLLM).

Usage

hc_list_llm_models(
  provider = "all",
  api_key = NULL,
  base_url = NULL,
  timeout_sec = 30
)

Arguments

Value

A data frame with columns provider, model, and info. Providers that cannot be reached (e.g. a missing API key) are skipped with a warning rather than aborting the whole call.

Examples

# Contacts the provider APIs; each provider needs its API key set in the
# environment (or a reachable server for "vllm").

if (nzchar(Sys.getenv("GEMINI_API_KEY"))) {
  hc_list_llm_models("gemini")
}
if (nzchar(Sys.getenv("OPENAI_API_KEY"))) {
  hc_list_llm_models("openai")
}

Description

Asks a large language model what biological program each module (or any other set of genes, proteins or other features) most likely represents. This is an AI-assisted interpretation, not a statistical test.

Usage

hc_llm_enrichment(
  hc = NULL,
  genes = NULL,
  modules = "all",
  system_prompt = NULL,
  context = NULL,
  provider = c("claude", "openai", "gemini", "local", "manual"),
  api_key = NULL,
  model = NULL,
  base_url = NULL,
  evidence = "enrichment",
  evidence_text = NULL,
  evidence_top = 10,
  evidence_qval = 0.05,
  rag_url = NULL,
  rag_top_k = 10,
  slot_name = "llm_enrichment",
  verbose = TRUE
)

Arguments

Details

Every set is interpreted in up to three separate requests, so that each answer sees only the input it is meant to see:

• general_processes: from the features alone, without context;

• contextual_state and key_regulators: from the features and the context;

• grounded_processes and supporting_evidence: from the features, the context and the evidence - by default the significant terms of hc_functional_enrichment(); optionally also the regulators of hc_upstream_inference() and literature passages from a DoRAG server.

hc_plot_llm_enrichment() draws one plot per field, so the plots can be compared directly: what the model makes of the genes alone, with context, and with statistical evidence. Without stored enrichment results the third request is left out (with a message).

The requests can be sent in two ways:

• through an API (provider = "claude", "openai", "gemini", or "local" for an OpenAI-compatible vLLM server), which needs an API key for the hosted providers;

• without an API key (provider = "manual"): one prompt per request type is prepared (the first is copied to the clipboard); paste each into a new chat of any provider and read every answer back with hc_llm_import(). The results are stored exactly like API results.

Results are stored in hc_satellite(hc, slot_name) (one entry per set) and hc_satellite(hc, paste0(slot_name, "_summary")) (a table).

Value

With hc: the updated HCoCenaExperiment (for "manual", with the pending request stored). Without hc: a data frame with one row per set (for "manual", an hc_llm_request object to pass to hc_llm_import()).

See Also

hc_llm_import(), hc_plot_llm_enrichment(), hc_list_llm_models()

Examples

# Without an API key: build the requests, answer them in a web chat, import.
req <- hc_llm_enrichment(
  genes = list(IFN = c("STAT1", "IRF7", "CXCL10", "GBP1", "IFI44L")),
  context = "Whole blood of patients with acute viral infection",
  evidence_text = list(IFN = "Enriched: type I interferon signaling (q = 1e-6)"),
  provider = "manual",
  verbose = FALSE
)
length(req$parts)  # genes only / with context / with evidence
cat(substr(req$parts[[1]], 1, 300))

# The answers copied from the web chat (here written by hand):
answer <- paste0(
  '{"request_id": "', req$id, '", "results": [{"set": "IFN", ',
  '"general_processes": "type I interferon signaling / antiviral response",',
  ' "contextual_state": "interferon-high antiviral state",',
  ' "key_regulators": "STAT1 / IRF7 / IRF9",',
  ' "grounded_processes": "type I interferon response",',
  ' "supporting_evidence": "type I interferon signaling"}]}'
)
hc_llm_import(req, text = answer, verbose = FALSE)


# Through an API (needs a key):
if (nzchar(Sys.getenv("ANTHROPIC_API_KEY"))) {
  hc_llm_enrichment(
    genes = c("STAT1", "IRF7", "CXCL10", "GBP1", "IFI44L"),
    provider = "claude"
  )
}

Description

Reads an answer to a request created with hc_llm_enrichment(provider = "manual"), checks that it belongs to that request, and stores it exactly like an API result. Markdown code fences, text around the JSON and typographic quotes are tolerated.

Usage

hc_llm_import(
  x,
  text = NULL,
  file = NULL,
  model = NULL,
  slot_name = "llm_enrichment",
  verbose = TRUE
)

Arguments

Details

A request has one part per request type (features only, with context, with evidence), and long requests are split further. Import every answer with this function; after each import the next open part is copied to the clipboard. An answer may also contain the fields of several parts.

Value

For an HCoCenaExperiment, the updated object. For an hc_llm_request, a data frame with one row per set answered so far.

See Also

hc_llm_enrichment()

Examples

req <- hc_llm_enrichment(
  genes = list(IFN = c("STAT1", "IRF7", "CXCL10")),
  provider = "manual", verbose = FALSE
)
answer <- paste0(
  "Here is the result:\n```json\n",
  '{"request_id": "', req$id, '", "results": [{"set": "IFN", ',
  '"general_processes": "interferon signaling", ',
  '"contextual_state": "antiviral state", "key_regulators": "STAT1"}]}',
  "\n```"
)
hc_llm_import(req, text = answer, model = "web chat", verbose = FALSE)

Description

Clusters donors based on module-time features produced by hc_longitudinal_module_means(). Candidate cluster numbers are scored via the Calinski-Harabasz index.

Usage

hc_longitudinal_endotype_clustering(
  hc,
  input_slot = "longitudinal_module_means",
  output_slot = "longitudinal_endotypes",
  k = 2:6,
  method = c("kmeans", "hclust"),
  scale_features = TRUE,
  nstart = 50,
  seed = 42,
  na_impute = c("median", "zero"),
  endotype_prefix = "E",
  module_k = NULL,
  module_method = c("kmeans", "hclust"),
  module_scale_features = TRUE,
  module_nstart = 30,
  module_seed = 42,
  module_na_impute = c("median", "zero")
)

Arguments

Value

Updated HCoCenaExperiment with endotype clustering in hc@satellite[[output_slot]].

Examples

hc <- hc_example_data("clustered")
hc <- hc_longitudinal_module_means(hc, donor_col = "donor", time_col = "timepoint")
hc <- hc_longitudinal_endotype_clustering(hc, k = 2)

Description

Clusters donors on CAP features (or alternative longitudinal feature matrices), providing CH-based model selection and PCA/UMAP embeddings.

Usage

hc_longitudinal_meta_clustering(
  hc,
  slot_name = "longitudinal_endotypes",
  output_slot = slot_name,
  feature_source = c("cap_matrix", "module_cluster_matrix", "feature_matrix_used"),
  k = 2:8,
  method = c("kmeans", "hclust"),
  scale_features = TRUE,
  nstart = 50,
  seed = 42,
  na_impute = c("median", "zero"),
  cluster_prefix = "MC",
  consensus = TRUE,
  consensus_runs = 200,
  consensus_sample_fraction = 0.8,
  consensus_feature_fraction = 0.8,
  consensus_linkage = "average",
  compute_umap = TRUE,
  umap_neighbors = 15,
  umap_min_dist = 0.3
)

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("clustered")
hc <- hc_longitudinal_step1_module_donor(
  hc,
  donor_col = "donor",
  time_col = "timepoint",
  time_levels = c("T1", "T2"),
  k = 2,
  nstart = 1,
  cap_runs = 1,
  impute = FALSE,
  seed = 1
)$hc
hc <- hc_longitudinal_endotype_clustering(hc, k = 2, nstart = 1, seed = 1)
hc <- hc_longitudinal_meta_clustering(
  hc,
  feature_source = "feature_matrix_used",
  k = 2:3,
  method = "kmeans",
  seed = 1
)

Description

CAP (cluster assignment proportions) quantifies for each donor the proportion of repeated clustering runs assigned to each module-specific cluster. This is the generalized, data-agnostic analogue of the CAP concept used in the legacy longitudinal scripts.

Usage

hc_longitudinal_module_cap(
  hc,
  slot_name = "longitudinal_endotypes",
  output_slot = slot_name,
  runs = 50,
  method = c("kmeans", "hclust"),
  nstart = 30,
  seed = 42,
  scale_features = TRUE,
  na_impute = c("median", "zero"),
  jitter_sd = 0
)

Arguments

Value

Updated HCoCenaExperiment with CAP outputs in hc@satellite[[output_slot]].

Examples

hc <- hc_example_data("clustered")
long <- hc_longitudinal_step1_module_donor(
  hc,
  donor_col = "donor",
  time_col = "timepoint",
  time_levels = c("T1", "T2"),
  k = 2,
  nstart = 1,
  cap_runs = 1,
  impute = FALSE,
  seed = 1
)
hc <- long$hc
hc <- hc_longitudinal_module_cap(hc, runs = 1)

Description

Computes module-level mean expression trajectories from the clustered modules in an HCoCenaExperiment. This creates a donor-by-(module x timepoint) feature matrix that can be used for longitudinal endotype discovery.

Usage

hc_longitudinal_module_means(
  hc,
  donor_col = "donor",
  time_col = "time_point",
  layer = NULL,
  group_col = NULL,
  modules = NULL,
  use_module_labels = TRUE,
  time_levels = NULL,
  aggregate_fun = function(x) base::mean(x, na.rm = TRUE),
  value_label = NULL,
  impute_missing = c("linear", "locf", "none"),
  min_genes_per_module = 5,
  slot_name = "longitudinal_module_means"
)

Arguments

Value

Updated HCoCenaExperiment with longitudinal data in hc@satellite[[slot_name]].

Examples

hc <- hc_example_data("clustered")
hc <- hc_longitudinal_module_means(hc, donor_col = "donor", time_col = "timepoint")

Description

Computes longitudinal module means, then clusters donors separately within each module directly on the donor-by-time trajectory matrix. The downstream quick steps remain unchanged and can be run via hc_longitudinal_step2_meta_clustering(), hc_longitudinal_step3_meta_module_trajectories(), and hc_plot_longitudinal_enrichment_meta_waves().

Usage

hc_longitudinal_step1_module_donor(
  hc,
  donor_col = "Subject",
  time_col = "Time_token",
  layer = NULL,
  time_levels,
  k = 2:6,
  method = c("kmeans", "hclust"),
  nstart = 50,
  cap_runs = 50,
  impute = TRUE,
  impute_method = "rfcont",
  ntree = 10,
  min_cluster_fraction = 0.1,
  score_method = c("calinski_harabasz", "tot_withinss"),
  scale_features = FALSE,
  na_impute = c("median", "zero"),
  cap_scale_features = scale_features,
  cap_na_impute = na_impute,
  cap_jitter_sd = 0,
  seed = 42,
  means_slot = "longitudinal_module_means",
  output_slot = "longitudinal_endotypes"
)

Arguments

Value

A list with updated hc, step plots, and diagnostics. When layer = "all" (or multiple layers are supplied), returns per-layer plot and diagnostic lists keyed by layer id.

Examples

hc <- hc_example_data("clustered")
res <- hc_longitudinal_step1_module_donor(
  hc,
  donor_col = "donor",
  time_col = "timepoint",
  time_levels = c("T1", "T2"),
  k = 2,
  nstart = 1,
  cap_runs = 1,
  impute = FALSE,
  seed = 1
)
hc <- res$hc

Description

Computes longitudinal module means, then clusters donors separately within each module directly on the donor-by-time trajectory matrix, especially for short trajectories with only a few timepoints.

Usage

hc_longitudinal_step1_module_donor_direct(
  hc,
  donor_col = "Subject",
  time_col = "Time_token",
  layer = NULL,
  time_levels,
  k = 2:6,
  method = c("kmeans", "hclust"),
  nstart = 50,
  cap_runs = 50,
  impute = TRUE,
  impute_method = "rfcont",
  ntree = 10,
  min_cluster_fraction = 0.1,
  score_method = c("calinski_harabasz", "tot_withinss"),
  scale_features = TRUE,
  na_impute = c("median", "zero"),
  cap_scale_features = scale_features,
  cap_na_impute = na_impute,
  cap_jitter_sd = 0,
  seed = 42,
  means_slot = "longitudinal_module_means_direct",
  output_slot = "longitudinal_endotypes_direct"
)

Arguments

Value

A list with updated hc, step plots, and diagnostics. When layer = "all" (or multiple layers are supplied), returns per-layer plot and diagnostic lists keyed by layer id.

Examples

hc <- hc_example_data("clustered")
res <- hc_longitudinal_step1_module_donor_direct(
  hc,
  donor_col = "donor",
  time_col = "timepoint",
  time_levels = c("T1", "T2"),
  k = 2,
  nstart = 1,
  cap_runs = 1,
  impute = FALSE,
  seed = 1
)
hc <- res$hc

Description

Runs the exact CAP -> PCA -> graph -> Leiden donor meta-clustering and returns the PCA/UMAP embeddings.

Usage

hc_longitudinal_step2_meta_clustering(
  hc,
  slot_name = "longitudinal_endotypes",
  dimensions = 4,
  graph_method = c("knn", "snn"),
  knn_method = c("annoy", "rann"),
  graph_k = 7,
  resolution = 0.4,
  leiden_method = "RBConfigurationVertexPartition",
  cluster_prefix = "MC",
  compute_umap = TRUE,
  umap_neighbors = 15,
  umap_min_dist = 0.3,
  seed = 42
)

Arguments

Value

A list with updated hc, step plots, and diagnostics. When multiple slots are processed, plots and diagnostics are returned as named lists keyed by layer/slot, plus a slot_info table describing the resolved slots.

Examples

hc <- hc_example_data("clustered")
hc <- hc_longitudinal_step1_module_donor(
  hc,
  donor_col = "donor",
  time_col = "timepoint",
  time_levels = c("T1", "T2"),
  k = 2,
  nstart = 1,
  cap_runs = 1,
  impute = FALSE,
  seed = 1
)$hc
res <- hc_longitudinal_step2_meta_clustering(hc, graph_k = 5, dimensions = 2)
hc <- res$hc

Description

Plots module trajectories after meta-clustering using donor and meta-cluster mean waves.

Usage

hc_longitudinal_step3_meta_module_trajectories(
  hc,
  slot_name = "longitudinal_endotypes",
  facet_ncol = 4,
  free_y = TRUE,
  square_panels = TRUE,
  value_mode = c("scaled_mean_vst", "expression"),
  value_range = c(-2, 2)
)

Arguments

Value

A list with unchanged hc and the step 3 plot. When multiple slots are processed, plots is returned as a named list keyed by layer/slot, plus a slot_info table describing the resolved slots.

Examples

hc <- hc_example_data("clustered")
hc <- hc_longitudinal_step1_module_donor(
  hc,
  donor_col = "donor",
  time_col = "timepoint",
  time_levels = c("T1", "T2"),
  k = 2,
  nstart = 1,
  cap_runs = 1,
  impute = FALSE,
  seed = 1
)$hc
hc <- hc_longitudinal_endotype_clustering(hc, k = 2, nstart = 1, seed = 1)
hc <- hc_longitudinal_meta_clustering(
  hc,
  feature_source = "feature_matrix_used",
  k = 2:3,
  method = "kmeans",
  consensus = TRUE,
  seed = 1
)
res <- hc_longitudinal_step3_meta_module_trajectories(hc)

Description

Runs the direct donor-by-time step 1 backend and then reuses the existing step 2 and step 3 quick wrappers on the resulting output slot family.

Usage

hc_longitudinal_workflow_direct(
  hc,
  donor_col = "Subject",
  time_col = "Time_token",
  layer = NULL,
  time_levels,
  k = 2:6,
  method = c("kmeans", "hclust"),
  nstart = 50,
  cap_runs = 50,
  impute = TRUE,
  impute_method = "rfcont",
  ntree = 10,
  min_cluster_fraction = 0.1,
  score_method = c("calinski_harabasz", "tot_withinss"),
  scale_features = TRUE,
  na_impute = c("median", "zero"),
  cap_scale_features = scale_features,
  cap_na_impute = na_impute,
  cap_jitter_sd = 0,
  seed = 42,
  means_slot = "longitudinal_module_means_direct",
  output_slot = "longitudinal_endotypes_direct"
)

Arguments

Value

A list with updated hc, nested plots, and per-step diagnostics.

Examples

hc <- hc_example_data("clustered")
res <- hc_longitudinal_workflow_direct(
  hc,
  donor_col = "donor",
  time_col = "timepoint",
  time_levels = c("T1", "T2"),
  k = 2,
  nstart = 1,
  cap_runs = 1,
  impute = FALSE,
  seed = 1
)
hc <- res$hc

Description

Access MultiAssayExperiment from an HCoCenaExperiment

Usage

hc_mae(x)

## S4 method for signature 'HCoCenaExperiment'
hc_mae(x)

Arguments

Value

A MultiAssayExperiment::MultiAssayExperiment object.

Examples

hc <- hc_init()
class(hc_mae(hc))

Description

Cuts the dendrogram of the module heatmap to merge modules with similar expression patterns. Run with save = FALSE first to preview a given k; save = TRUE overwrites the current module assignment and cannot be undone.

Usage

hc_merge_clusters(
  hc,
  k = "auto",
  save = TRUE,
  method = "complete",
  k_min = 2,
  k_max = NULL,
  auto_parsimony_penalty = 1e-04,
  verbose = TRUE
)

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("clustered")
hc <- hc_plot_cluster_heatmap(hc, file_name = FALSE)
hc <- hc_merge_clusters(hc, k = 2, save = FALSE)

Description

Correlate categorical metadata with modules (S4 API)

Usage

hc_meta_correlation_cat(hc, meta, set, p_val = 0.05, padj = "BH")

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("clustered")
hc_meta_correlation_cat(hc, set = 1, meta = "batch")

Description

Correlates the mean expression of each module's genes per sample with one or more numeric annotation columns, and shows the result as a heatmap. Non-significant correlations are left blank.

Usage

hc_meta_correlation_num(hc, set, meta, p_val = 0.05, padj = "BH")

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("clustered")
hc_meta_correlation_num(hc, set = 1, meta = "age")

Description

Meta-data plotting (S4 API)

Usage

hc_meta_plot(
  hc,
  set,
  group_col = NULL,
  meta_col = NULL,
  type = "cat",
  cols = NULL
)

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("prepared")
hc <- hc_meta_plot(hc, set = 1, group_col = "group", meta_col = "batch")

Description

Test module differences between conditions (S4 API)

Usage

hc_module_condition_significance(
  hc,
  set = "all",
  condition_col = NULL,
  donor_col = NULL,
  time_col = NULL,
  run_wilcox = TRUE,
  run_limma = TRUE,
  run_lmm = FALSE,
  lmm_include_time = TRUE,
  limma_reference = NULL,
  limma_trend = TRUE,
  standardize_modules = FALSE,
  padj = "BH",
  export_excel = TRUE,
  excel_file = "Module_condition_significance.xlsx",
  slot_name = "module_condition_significance"
)

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("clustered")
hc <- hc_module_condition_significance(
  hc,
  condition_col = "group",
  run_limma = FALSE,
  export_excel = FALSE
)

Description

Returns the Group-Fold-Change values behind the module heatmap as tables: once aggregated per group (the numbers shown in the heatmap) and once resolved per individual sample.

Usage

hc_module_gfc_tables(
  hc,
  layer = 1,
  group_col = NULL,
  modules = "all",
  annotate = TRUE,
  export_xlsx = TRUE,
  out_dir = NULL,
  file_name = "Module_GFC_group_and_sample.xlsx",
  verbose = TRUE
)

Arguments

Details

hCoCena itself only stores the group level (module_gfc_means, modules by groups). This function computes the per-sample counterpart using hCoCena's own GFC definition, so both tables live on the same scale:

1. when data_in_log is TRUE, values are anti-logged (base 2) and all averaging happens on the linear scale,

2. group means (or the raw sample values) are formed per gene,

3. the reference follows the control_keyword global setting, exactly as hc_run_expression_analysis_2() does. With "none" it is the unweighted mean of the group means – not the mean across all samples, which differs whenever the groups are unequally sized. With a control keyword it is the mean of the matching control group, and that group is dropped from the group table just like in the heatmap,

4. foldchange(x, ref) is x / ref when x >= ref and -ref / x otherwise, a signed ratio rather than a difference of logs,

5. values are clamped to ⁠+/- range_GFC⁠,

6. the module value is the mean across the genes of that module.

Because step 4 is non-linear and step 5 truncates, the group table is not the average of the sample table (fc(mean(x)) != mean(fc(x))). Both are correct but answer different questions: by_group is the value plotted in the heatmap, by_sample shows how uniformly the samples carry it. Use by_sample for sample-level statistics and by_group for figure labels.

As a self-check the function compares its own by_group against the grp_means stored by hCoCena and reports the largest absolute difference, which is expected to be on the order of 1e-3 – the rounding hCoCena applies per gene.

Value

Updated HCoCenaExperiment. The tables are stored in hc@satellite$module_gfc_tables as by_group, by_group_long, by_sample, by_sample_long, plus reference and settings.

See Also

hc_plot_cluster_heatmap(), which produces the group-level module_gfc_means table.

Examples

hc <- hc_example_data("clustered")
hc <- hc_module_gfc_tables(hc, export_xlsx = FALSE)
hc_satellite(hc, "module_gfc_tables")$by_group

Description

Compares two externally stored gene-to-cluster tables and visualises how the modules of the first assignment map onto those of the second.

Usage

hc_network_comparison_1(
  hc,
  gtc1_path,
  gtc2_path,
  sep = "\t",
  header = TRUE,
  cellsize = 18
)

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("clustered")
gtc <- hc_gene_to_cluster(hc)
f1 <- tempfile(fileext = ".tsv")
f2 <- tempfile(fileext = ".tsv")
utils::write.table(gtc, f1, sep = "\t", row.names = FALSE, quote = FALSE)
utils::write.table(gtc, f2, sep = "\t", row.names = FALSE, quote = FALSE)
hc <- hc_network_comparison_1(hc, gtc1_path = f1, gtc2_path = f2)

Description

Contrasts the neighbourhoods of gene_vec between two networks and reports which edges are shared and which are unique to either network.

Usage

hc_network_comparison_2(hc, net1, net2, as = "igraph", gene_vec)

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("clustered")
g <- hc_graph(hc)
hc <- hc_network_comparison_2(
  hc,
  net1 = g,
  net2 = g,
  gene_vec = igraph::V(g)$name[1:3]
)

Description

PCA plotting (S4 API)

Usage

hc_pca(hc, which = "all", color_by = NULL, ellipses = FALSE, cols = NULL)

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("prepared")
hc <- hc_pca(hc)

Description

PCA algorithm comparison (S4 API)

Usage

hc_pca_algo_compare(hc, gtc = NULL, algo = NULL, cols = NULL)

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("clustered")
hc <- hc_pca_algo_compare(hc)

Description

Uses hc@satellite$celltype_annotation and draws a heatmap-like matrix (modules x cell types) for easier interpretation than stacked barplots.

Usage

hc_plot_celltype_annotation_matrix(
  hc,
  database = NULL,
  value = c("pct", "count"),
  top_celltypes = 15,
  include_other = TRUE,
  other_label = "Other",
  cluster_order = NULL,
  low_color = "#f7fbff",
  high_color = "#08306b",
  return_data = TRUE
)

Arguments

Value

A ggplot object or a list containing plot + data.

Examples

hc <- hc_example_data("clustered")
gmt <- system.file("extdata", "toy_celltype_markers.gmt", package = "hcocena")
hc <- hc_celltype_annotation(
  hc,
  databases = character(0),
  custom_gmt_files = c(CellTypes = gmt),
  export_excel = FALSE
)
p <- hc_plot_celltype_annotation_matrix(hc)

Description

Plot module heatmap (S4 API)

Usage

hc_plot_cluster_heatmap(hc, file_name = "Heatmap_modules.pdf", ...)

Arguments

Value

Updated HCoCenaExperiment.

Examples

hc <- hc_example_data("clustered")
hc <- hc_plot_cluster_heatmap(hc, file_name = FALSE)

Description

Creates a module heatmap view for selected modules in the exact order supplied by modules. The module labels and module-color annotations are preserved, and the GFC expression color scale is inherited from the full heatmap cache instead of being recomputed from the subset.

Usage

hc_plot_cluster_heatmap_view(
  hc,
  modules,
  file_name = "Heatmap_modules_view.pdf",
  col_order = NULL,
  cluster_columns = FALSE,
  cluster_rows = FALSE,
  gfc_colors = NULL,
  gfc_scale_limits = NULL,
  preserve_full_heatmap = TRUE,
  store_view = TRUE,
  view_name = NULL,
  ...
)