Package 'nichenetr'

Description

add_hyperparameters_parameter_settings will generate a list of lists containing the parameter values that need to be used for model construction.

Usage

add_hyperparameters_parameter_settings(parameter_setting, lr_sig_hub, gr_hub, ltf_cutoff, algorithm, damping_factor, correct_topology)

Arguments

Value

A list of lists. Every sub-list contains parameter values for a different parameter.

Examples

## Not run: 
weights_settings_loi <- prepare_settings_leave_one_in_characterization(lr_network, sig_network, gr_network, source_weights_df)
weights_settings_loi <- lapply(weights_settings_loi, add_hyperparameters_parameter_settings, lr_sig_hub = 0.25, gr_hub = 0.5, ltf_cutoff = 0, algorithm = "PPR", damping_factor = 0.8, correct_topology = TRUE)

## End(Not run)

Description

add_ligand_popularity_measures_to_perfs: Get a data.frame in which the performance measures for target gene prediction of a ligand are merged with the number of times the ligand is mentioned in the Pubmed literature. Serves to investigate popularity bias (i.e. it can be expected that frequenly cited ligands will have better predictive performance because they are better studied).

Usage

add_ligand_popularity_measures_to_perfs(performances, ncitations)

Arguments

Value

A data.frame in which the performance measures for target gene prediction of a ligand are merged with the number of times the ligand is mentioned in the Pubmed literature.

Examples

## Not run: 
library(dplyr)
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
settings <- lapply(expression_settings_validation[1:10], convert_expression_settings_evaluation)
ligands <- extract_ligands_from_settings(settings)
ligand_target_matrix <- construct_ligand_target_matrix(weighted_networks, lr_network, ligands)
performances <- bind_rows(lapply(settings, evaluate_target_prediction, ligand_target_matrix))
# ncitations = get_ncitations_genes()
performances_ligand_popularity <- add_ligand_popularity_measures_to_perfs(performances, ncitations)

## End(Not run)

Description

add_new_datasource adds a new data source to one of the ligand-receptor, signaling and gene regulatory data sources.

Usage

add_new_datasource(new_source, network, new_weight,source_weights_df)

Arguments

Value

A list containing 2 elements (network and source_weights_df): the updated network containing your data source and the updated source_weights_df containing the weight of the newly added data source.

Examples

## Not run: 
## Update the lr_network with a new ligand-receptor data source
library(dplyr)
lr_toy <- tibble(from = "A", to = "B", source = "toy")
new_lr_network <- add_new_datasource(lr_toy, lr_network, 1, source_weights_df)

## End(Not run)

Description

alias_to_symbol_seurat Convert aliases to official gene symbols in a Seurat Object. Makes use of 'convert_alias_to_symbols'

Usage

alias_to_symbol_seurat(seurat_obj, organism)

Arguments

Value

Seurat object

Examples

## Not run: 
seurat_object_lite <- readRDS(url("https://zenodo.org/record/3531889/files/seuratObj_test.rds"))
seurat_object_lite <- seurat_object_lite %>% alias_to_symbol_seurat("human")

## End(Not run)

Description

A data.frame/tibble describing for each type of data source the "parent" database and type of interactions it countains.

Usage

annotation_data_sources

Format

A data frame/tibble

source

Name of the data source

database

Name of the comprehensive database to which the data source belongs

type_db

Type of data source

type_interaction

Interaction type

network

Network layer to which the data source belongs: ligand_receptor, signaling or gene_regulatory

Description

apply_hub_corrections downweighs the importance of nodes with a lot of incoming links in the ligand-signaling and/or gene regulatory network. Hub correction method according to following equation: $Wcor =W * D^-h$ with $D$ the indegree matrix of the respective network and $h$ the correction factor.

Usage

apply_hub_corrections(weighted_networks, lr_sig_hub, gr_hub)

Arguments

Value

A list containing 2 elements (lr_sig and gr): the hubiness-corrected integrated weighted ligand-signaling and gene regulatory networks in data frame / tibble format with columns: from, to, weight.

Examples

## Not run: 
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
wn <- apply_hub_corrections(weighted_networks, lr_sig_hub = 0.5, gr_hub = 0.5)

## End(Not run)

Description

assess_influence_source will assess the influence of an individual data source on ligand-target probability scores (or rankings of these). Possible output: the ligand-target matrices of the complete model vs the leave-one-out model in which the data source of interest was left out; or a list indicating which target genes for every ligand of interest are affected the most by leaving out the data source of interest.

Usage

assess_influence_source(source, lr_network, sig_network, gr_network, source_weights_df, ligands, rankings = FALSE, matrix_output = FALSE,  secondary_targets = FALSE, remove_direct_links = "no", ...)

Arguments

Value

If matrix_output == TRUE: A list of sublists; every sublist contains the elements $model_name and $model: the constructed ligand-target matrix. If matrix_output == FALSE: A list of sublist: every sublist contains; $ligand: name of the ligand tested; $targets_higher: sorted vector of ligand-target scores or rankings of target that score higher in the complete model compared to the leave-one-out model; targets_lower: sorted vector of ligand-target scores or rankings of target that score lower in the complete model compared to the leave-one-out model.

Examples

## Not run: 
ligands <- extract_ligands_from_settings(expression_settings_validation[1:4])
output <- assess_influence_source("ontogenet", lr_network, sig_network, gr_network, source_weights_df, ligands, lr_sig_hub = 0.25, gr_hub = 0.5, ltf_cutoff = 0, algorithm = "PPR", damping_factor = 0.8, correct_topology = TRUE)

## End(Not run)

Description

assess_rf_class_probabilities Assess probability that a target gene belongs to the geneset based on a multi-ligand random forest model (with cross-validation). Target genes and background genes will be split in different groups in a stratified way.

Usage

assess_rf_class_probabilities(round,folds,geneset,background_expressed_genes,ligands_oi,ligand_target_matrix)

Arguments

Value

A tibble with columns: $gene, $response, $prediction. Response indicates whether the gene belongs to the geneset of interest, prediction gives the probability this gene belongs to the geneset according to the random forest model.

Examples

## Not run: 
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
ligands <- list("TNF", "BMP2", "IL4")
ligand_target_matrix <- construct_ligand_target_matrix(weighted_networks, lr_network, ligands, ltf_cutoff = 0, algorithm = "PPR", damping_factor = 0.5, secondary_targets = FALSE)
potential_ligands <- c("TNF", "BMP2", "IL4")
geneset <- c("SOCS2", "SOCS3", "IRF1")
background_expressed_genes <- c("SOCS2", "SOCS3", "IRF1", "ICAM1", "ID1", "ID2", "ID3")
fold1_rf_prob <- assess_rf_class_probabilities(round = 1, folds = 2, geneset = geneset, background_expressed_genes = background_expressed_genes, ligands_oi = potential_ligands, ligand_target_matrix = ligand_target_matrix)

## End(Not run)

Description

Assign ligands to a sender cell type, based on the strongest expressing cell type of that ligand. Ligands are only assigned to a cell type if that cell type is the only one to show an expression that is higher than the average + SD. Otherwise, it is assigned to "General".

Usage

assign_ligands_to_celltype(
  seuratObj,
  ligands,
  celltype_col,
  func.agg = mean,
  func.assign = function(x) {
     mean(x) + sd(x)
 },
  condition_oi = NULL,
  condition_col = NULL,
  ...
)

Arguments

Details

If the provided slot/layer is "data", the normalized counts are first exponentiated before aggregation is performed

Value

A data frame of two columns, the cell type the ligand has been assigned to (ligand_type) and the ligand name (ligand)

Examples

## Not run: 
assign_ligands_to_celltype(
  seuratObj = seuratObj, ligands = best_upstream_ligands[1:20],
  celltype_col = "celltype", func.agg = mean, func.assign = function(x) {
    mean(x) + sd(x)
  },
  condition_oi = "LCMV", condition_col = "aggregate", slot = "data"
)

## End(Not run)

Description

bootstrap_ligand_activity_analysis Randomly sample a gene set from all expressed genes in the receiver cell type, then perform ligand activity analysis on this gene set. This 'null gene set' has equal length to the gene set of interest.

Usage

bootstrap_ligand_activity_analysis(expressed_genes_receiver, geneset_oi, background_expressed_genes, ligand_target_matrix, potential_ligands, n_iter = 10, n_cores = 1, parallel_func = "mclapply")

Arguments

Value

List of n_iter elements, each element containing the output of predict_ligand_activities for a random gene set

Examples

## Not run: 
permutations <- bootstrap_ligand_activity_analysis(expressed_genes_receiver, geneset_oi, background_expressed_genes,
  ligand_target_matrix, potential_ligands,
  n_iter = 10, n_cores = 1, parallel_func = "mclapply"
)

## End(Not run)

Description

calculate_de Calculate differential expression of one cell type versus all other cell types using Seurat::FindAllMarkers. If condition_oi is provided, only consider cells from that condition.

Usage

calculate_de(seurat_obj, celltype_colname, condition_oi = NA, condition_colname = NA, assay_oi = "RNA", ...)

Arguments

Value

A dataframe containing the DE results

Examples

## Not run: 
seurat_obj <- readRDS(url("https://zenodo.org/record/3531889/files/seuratObj.rds"))
seurat_obj$celltype <- make.names(seurat_obj$celltype)
# Calculate cell-type specific markers across conditions
calculate_de(seurat_obj, "celltype")
# Calculate LCMV-specific cell-type markers
calculate_de(seurat_obj, "celltype", condition_oi = "LCMV", condition_colname = "aggregate")

## End(Not run)

Description

calculate_fraction_top_predicted Defines the fraction of genes belonging to the geneset or background and to the top-predicted genes.

Usage

calculate_fraction_top_predicted(affected_gene_predictions, quantile_cutoff = 0.95)

Arguments

Value

A tibble indicating the number of genes belonging to the gene set of interest or background (true_target column), the number and fraction of genes of these gruops that were part of the top predicted targets in a specific cross-validation round.

Description

calculate_fraction_top_predicted_fisher Performs a Fisher's exact test to determine whether genes belonging to the gene set of interest are more likely to be part of the top-predicted targets.

Usage

calculate_fraction_top_predicted_fisher(affected_gene_predictions, quantile_cutoff = 0.95, p_value_output = TRUE)

Arguments

Value

Summary of the Fisher's exact test or just the p-value

Examples

## Not run: 
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
ligands <- list("TNF", "BMP2", "IL4")
ligand_target_matrix <- construct_ligand_target_matrix(weighted_networks, lr_network, ligands, ltf_cutoff = 0, algorithm = "PPR", damping_factor = 0.5, secondary_targets = FALSE)
potential_ligands <- c("TNF", "BMP2", "IL4")
geneset <- c("SOCS2", "SOCS3", "IRF1")
background_expressed_genes <- c("SOCS2", "SOCS3", "IRF1", "ICAM1", "ID1", "ID2", "ID3")
gene_predictions_list <- seq(2) %>% lapply(assess_rf_class_probabilities, 2, geneset = geneset, background_expressed_genes = background_expressed_genes, ligands_oi = potential_ligands, ligand_target_matrix = ligand_target_matrix)
target_prediction_performances_fisher_pval <- gene_predictions_list %>%
  lapply(calculate_fraction_top_predicted_fisher) %>%
  unlist() %>%
  mean()

## End(Not run)

Description

calculate_niche_de Calculate differential expression of cell types in one niche versus all other niches of interest. This is possible for sender cell types and receiver cell types.

Usage

calculate_niche_de(seurat_obj, niches, type, assay_oi = "SCT")

Arguments

Value

A tibble containing the DE results of the niches versus each other.

Examples

## Not run: 
seurat_obj <- readRDS(url("https://zenodo.org/record/5840787/files/seurat_obj_subset_integrated_zonation.rds"))
niches <- list(
  "KC_niche" = list(
    "sender" = c("LSECs_portal", "Hepatocytes_portal", "Stellate cells_portal"),
    "receiver" = c("KCs")
  ),
  "MoMac2_niche" = list(
    "sender" = c("Cholangiocytes", "Fibroblast 2"),
    "receiver" = c("MoMac2")
  ),
  "MoMac1_niche" = list(
    "sender" = c("Capsule fibroblasts", "Mesothelial cells"),
    "receiver" = c("MoMac1")
  )
)
calculate_niche_de(seurat_obj, niches, "sender")

## End(Not run)

Description

calculate_niche_de_targets Calculate differential expression of receiver cell type in one niche versus all other niches of interest: focus on finding DE genes

Usage

calculate_niche_de_targets(seurat_obj, niches, expression_pct, lfc_cutoff, assay_oi = "SCT")

Arguments

Value

A tibble containing the DE results of the niches versus each other.

Examples

## Not run: 
seurat_obj <- readRDS(url("https://zenodo.org/record/5840787/files/seurat_obj_subset_integrated_zonation.rds"))
niches <- list(
  "KC_niche" = list(
    "sender" = c("LSECs_portal", "Hepatocytes_portal", "Stellate cells_portal"),
    "receiver" = c("KCs")
  ),
  "MoMac2_niche" = list(
    "sender" = c("Cholangiocytes", "Fibroblast 2"),
    "receiver" = c("MoMac2")
  ),
  "MoMac1_niche" = list(
    "sender" = c("Capsule fibroblasts", "Mesothelial cells"),
    "receiver" = c("MoMac1")
  )
)
calculate_niche_de_targets(seurat_obj, niches, expression_pct = 0.10, lfc_cutoff = 0.15)

## End(Not run)

Description

calculate_p_value_bootstrap Calculate the p-value for each ligand from the bootstrapped ligand activity analysis

Usage

calculate_p_value_bootstrap(bootstrap_results, ligand_activities, metric = "aupr_corrected")

Arguments

Value

Data frame with the ligand name, the number of times the bootstrapped value had a higher score than the observed (total), and the p-value for each ligand, calculated as total/length(bootstrap_results)

Examples

## Not run: 
permutations <- bootstrap_ligand_activity_analysis(expressed_genes_receiver, geneset_oi, background_expressed_genes,
  ligand_target_matrix, potential_ligands,
  n_iter = 10
)
p_values <- calculate_p_value_bootstrap(permutations, ligand_activities, metric = "aupr_corrected")

## End(Not run)

Description

calculate_spatial_DE Calculate differential expression between spatially different subpopulations of the same cell type

Usage

calculate_spatial_DE(seurat_obj, spatial_info, assay_oi = "SCT")

Arguments

Value

A tibble with DE output

Examples

## Not run: 
seurat_obj <- readRDS(url("https://zenodo.org/record/5840787/files/seurat_obj_subset_integrated_zonation.rds"))
spatial_info <- tibble(
  celltype_region_oi = c("LSECs_portal", "Hepatocytes_portal", "Stellate cells_portal"),
  celltype_other_region = c("LSECs_central", "Hepatocytes_central", "Stellate cells_central")
) %>%
  mutate(niche = "KC_niche", celltype_type = "sender")
calculate_spatial_DE(seurat_obj, spatial_info)

## End(Not run)

Description

classification_evaluation_continuous_pred_wrapper Assess how well classification predictions accord to the expected response.

Usage

classification_evaluation_continuous_pred_wrapper(response_prediction_tibble)

Arguments

Value

A tibble showing several classification evaluation metrics.

Examples

## Not run: 
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
ligands <- list("TNF", "BMP2", "IL4")
ligand_target_matrix <- construct_ligand_target_matrix(weighted_networks, lr_network, ligands, ltf_cutoff = 0, algorithm = "PPR", damping_factor = 0.5, secondary_targets = FALSE)
potential_ligands <- c("TNF", "BMP2", "IL4")
geneset <- c("SOCS2", "SOCS3", "IRF1")
background_expressed_genes <- c("SOCS2", "SOCS3", "IRF1", "ICAM1", "ID1", "ID2", "ID3")
fold1_rf_prob <- assess_rf_class_probabilities(round = 1, folds = 2, geneset = geneset, background_expressed_genes = background_expressed_genes, ligands_oi = potential_ligands, ligand_target_matrix = ligand_target_matrix)
classification_evaluation_continuous_pred_wrapper(fold1_rf_prob)

## End(Not run)

Description

Removes all cached databases from memory.

Usage

clear_database_cache()

Description

combine_sender_receiver_de Combine the differential expression information of ligands in the sender celltypes with the differential expression information of their cognate receptors in the receiver cell types.

Usage

combine_sender_receiver_de(DE_sender_processed, DE_receiver_processed, lr_network, specificity_score = "min_lfc")

Arguments

Value

A tibble giving the differential expression information of ligands in the sender celltypes and their cognate receptors in the receiver cell types.

Examples

## Not run: 
seurat_obj <- readRDS(url("https://zenodo.org/record/5840787/files/seurat_obj_subset_integrated_zonation.rds"))
niches <- list(
  "KC_niche" = list(
    "sender" = c("LSECs_portal", "Hepatocytes_portal", "Stellate cells_portal"),
    "receiver" = c("KCs")
  ),
  "MoMac2_niche" = list(
    "sender" = c("Cholangiocytes", "Fibroblast 2"),
    "receiver" = c("MoMac2")
  ),
  "MoMac1_niche" = list(
    "sender" = c("Capsule fibroblasts", "Mesothelial cells"),
    "receiver" = c("MoMac1")
  )
)
DE_sender <- calculate_niche_de(seurat_obj, niches, "sender")
DE_receiver <- calculate_niche_de(seurat_obj, niches, "receiver")
expression_pct <- 0.10
DE_sender_processed <- process_niche_de(DE_table = DE_sender, niches = niches, expression_pct = expression_pct, type = "sender")
DE_receiver_processed <- process_niche_de(DE_table = DE_receiver, niches = niches, expression_pct = expression_pct, type = "receiver")
specificity_score_LR_pairs <- "min_lfc"
DE_sender_receiver <- combine_sender_receiver_de(DE_sender_processed, DE_receiver_processed, lr_network, specificity_score = specificity_score_LR_pairs)

## End(Not run)

Description

construct_ligand_target_matrix Convert integrated weighted networks into a matrix containg ligand-target probability scores. The higher this score, the more likely a particular ligand can induce the expression of a particular target gene.

Usage

construct_ligand_target_matrix(weighted_networks, lr_network, ligands, ltf_cutoff = 0.99, algorithm = "PPR", damping_factor = 0.5, secondary_targets = FALSE,ligands_as_cols = TRUE, remove_direct_links = "no")

Arguments

Value

A matrix containing ligand-target probability scores.

Examples

## Not run: 
## Generate the ligand-target matrix from loaded weighted_networks
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
ligands <- list("TNF", "BMP2", c("IL4", "IL13"))
ligand_target_matrix <- construct_ligand_target_matrix(weighted_networks, lr_network, ligands, ltf_cutoff = 0.99, algorithm = "PPR", damping_factor = 0.5, secondary_targets = FALSE, remove_direct_links = "no")

## End(Not run)

Description

construct_ligand_tf_matrix Convert integrated weighted networks into a matrix containg ligand-tf probability scores. The higher this score, the more likely a particular ligand can signal to a downstream gene.

Usage

construct_ligand_tf_matrix(weighted_networks, lr_network, ligands, ltf_cutoff = 0.99, algorithm = "PPR", damping_factor = 0.5, ligands_as_cols = FALSE)

Arguments

Value

A matrix containing ligand-target probability scores.

Examples

## Not run: 
## Generate the ligand-target matrix from loaded weighted_networks
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
ligands <- list("TNF", "BMP2", c("IL4", "IL13"))
ligand_tf <- construct_ligand_tf_matrix(weighted_networks, lr_network, ligands, ltf_cutoff = 0.99, algorithm = "PPR", damping_factor = 0.5, ligands_as_cols = TRUE)

## End(Not run)

Description

construct_model will take as input a setting of parameters (data source weights and hyperparameters) and layer-specific networks to construct a ligand-target matrix.

Usage

construct_model(parameters_setting, lr_network, sig_network, gr_network, ligands, secondary_targets = FALSE, remove_direct_links = "no")

Arguments

Value

A list containing following elements: $model_name and $model.

Examples

## Not run: 
library(dplyr)
settings <- lapply(expression_settings_validation[1:4], convert_expression_settings_evaluation)
weights_settings_loi <- prepare_settings_leave_one_in_characterization(lr_network, sig_network, gr_network, source_weights_df)
weights_settings_loi <- lapply(weights_settings_loi, add_hyperparameters_parameter_settings, lr_sig_hub = 0.25, gr_hub = 0.5, ltf_cutoff = 0, algorithm = "PPR", damping_factor = 0.8, correct_topology = TRUE)
doMC::registerDoMC(cores = 8)
ligands <- extract_ligands_from_settings(settings)
models_characterization <- parallel::mclapply(weights_settings_loi[1:3], construct_model, lr_network, sig_network, gr_network, ligands, mc.cores = 3)

## End(Not run)

Description

construct_random_model will take as input a setting of parameters (data source weights and hyperparameters) and layer-specific networks to construct a ligand-target matrix after randomization by edge swapping.

Usage

construct_random_model(parameters_setting, lr_network, sig_network, gr_network, ligands, secondary_targets = FALSE, remove_direct_links = "no")

Arguments

Value

A list containing following elements: $model_name and $model.

Examples

## Not run: 
library(dplyr)
settings <- lapply(expression_settings_validation[1:4], convert_expression_settings_evaluation)
weights_settings_loi <- prepare_settings_leave_one_in_characterization(lr_network, sig_network, gr_network, source_weights_df)
weights_settings_loi <- lapply(weights_settings_loi, add_hyperparameters_parameter_settings, lr_sig_hub = 0.25, gr_hub = 0.5, ltf_cutoff = 0, algorithm = "PPR", damping_factor = 0.8, correct_topology = TRUE)
doMC::registerDoMC(cores = 8)
ligands <- extract_ligands_from_settings(settings)
models_characterization <- parallel::mclapply(weights_settings_loi[1:3], construct_random_model, lr_network, sig_network, gr_network, ligands, mc.cores = 3)

## End(Not run)

Description

construct_tf_target_matrix Convert integrated gene regulatory weighted network into matrix format.

Usage

construct_tf_target_matrix(weighted_networks, tfs_as_cols = FALSE, standalone_output = FALSE)

Arguments

Value

A matrix containing tf-target regulatory weights.

Examples

## Not run: 
## Generate the ligand-target matrix from loaded weighted_networks
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
tf_target <- construct_tf_target_matrix(weighted_networks, tfs_as_cols = TRUE, standalone_output = TRUE)

## End(Not run)

Description

construct_weighted_networks construct layer-specific weighted integrated networks from input source networks via weighted aggregation.

Usage

construct_weighted_networks(lr_network, sig_network, gr_network,source_weights_df, n_output_networks = 2)

Arguments

Value

A list containing 2 elements (lr_sig and gr) or 3 elements (lr, sig, gr): the integrated weighted ligand-signaling and gene regulatory networks or ligand-receptor, signaling and gene regulatory networks in data frame / tibble format with columns: from, to, weight.

Examples

## Not run: 
## Generate the weighted networks from input source networks
wn <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)

## End(Not run)

Description

convert_alias_to_symbols Convert aliases to offcial gene symbols

Usage

convert_alias_to_symbols(aliases, organism, verbose = TRUE)

Arguments

Value

A character vector of official gene symbols.

Examples

library(dplyr)
human_symbols <- c("TNF", "IFNG", "IL-15")
aliases <- human_symbols %>% convert_alias_to_symbols(organism = "human")

Description

convert_cluster_to_settings Convert cluster assignment to settings format suitable for target gene prediction.

Usage

convert_cluster_to_settings(i, cluster_vector, setting_name, setting_from, background = NULL)

Arguments

Value

A list with following elements: $name (indicating the cluster id), $from, $response. $response is a gene-named logical vector indicating whether the gene is part of the respective cluster.

Examples

## Not run: 
genes_clusters <- c("TGFB1" = 1, "TGFB2" = 1, "TGFB3" = 2)
cluster_settings <- lapply(seq(length(unique(genes_clusters))), convert_cluster_to_settings, cluster_vector = genes_clusters, setting_name = "example", setting_from = "BMP2")

## End(Not run)

Description

convert_expression_settings_evaluation Converts expression settings to correct settings format for evaluation of target gene prediction.

Usage

convert_expression_settings_evaluation(setting)

Arguments

Value

A list with following elements: $name, $from, $response. $response will be a gene-named logical vector indicating whether the gene's transcription was influenced by the active ligand(s) in the setting of interest.

Examples

## Not run: 
settings <- lapply(expression_settings_validation, convert_expression_settings_evaluation)

## End(Not run)

Description

convert_expression_settings_evaluation_regression Converts expression settings to correct settings format for evaluation of target gene log fold change prediction (regression).

Usage

convert_expression_settings_evaluation_regression(setting)

Arguments

Value

A list with following elements: $name, $from, $response. $response will be a gene-named numeric vector of log fold change values.

Examples

## Not run: 
settings <- lapply(expression_settings_validation, convert_expression_settings_evaluation_regression)

## End(Not run)

Description

convert_gene_list_settings_evaluation Converts a gene list to correct settings format for evaluation of target gene prediction.

Usage

convert_gene_list_settings_evaluation(gene_list, name, ligands_oi, background)

Arguments

Value

A list with following elements: $name, $from, $response

Examples

## Not run: 
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
all_genes <- unique(c(weighted_networks$gr$from, weighted_networks$gr$to, weighted_networks$lr_sig$from, weighted_networks$lr_sig$to))
gene_list <- c("ID1", "ID2", "ID3")
setting <- list(convert_gene_list_settings_evaluation(gene_list = c("ID1", "ID2", "ID3"), name = "test", ligands_oi = "TGFB1", background = all_genes))

## End(Not run)

Description

convert_human_to_mouse_symbols Convert human gene symbols to their mouse one-to-one orthologs.

Usage

convert_human_to_mouse_symbols(symbols, version = 1)

Arguments

Value

A character vector of official mouse gene symbols (one-to-one orthologs of the input human gene symbols).

Examples

library(dplyr)
human_symbols <- c("TNF", "IFNG")
mouse_symbols <- human_symbols %>% convert_human_to_mouse_symbols()

Description

convert_mouse_to_human_symbols Convert mouse gene symbols to their human one-to-one orthologs.

Usage

convert_mouse_to_human_symbols(symbols, version = 1)

Arguments

Value

A character vector of official human gene symbols (one-to-one orthologs of the input mouse gene symbols).

Examples

library(dplyr)
mouse_symbols <- c("Tnf", "Ifng")
human_symbols <- mouse_symbols %>% convert_mouse_to_human_symbols()

Description

convert_settings_ligand_prediction Converts settings to correct settings format for ligand activity prediction. In this prediction problem, ligands (out of a set of possibly active ligands) will be ranked based on feature importance scores. The format can be made suited for: 1) validation of ligand activity state prediction by calculating individual feature importane scores or 2) feature importance based on models with embedded feature importance determination; applications in which ligands need to be scores based on their possible upstream activity: 3) by calculating individual feature importane scores or 4) feature importance based on models with embedded feature importance determination.

Usage

convert_settings_ligand_prediction(settings, all_ligands, validation = TRUE, single = TRUE)

Arguments

Value

A list with following elements: $name, $ligand: name of active ligand(s) (only if validation is TRUE), $from (ligand(s) that will be tested for activity prediction), $response

Examples

## Not run: 
settings <- lapply(expression_settings_validation, convert_expression_settings_evaluation)
ligands <- unlist(extract_ligands_from_settings(settings, combination = FALSE))
settings_ligand_pred <- convert_settings_ligand_prediction(settings, ligands, validation = TRUE, single = TRUE)

## End(Not run)

Description

convert_settings_tf_prediction Converts settings to correct settings format for TF activity prediction. In this prediction problem, TFs (out of a set of possibly active TFs) will be ranked based on feature importance scores. The format can be made suited for applications in which TFs need to be scored based on their possible upstream activity: 3) by calculating individual feature importane scores or 4) feature importance based on models with embedded feature importance determination. Remark that upstream regulator analysis for TFs here is experimental and was not thoroughly validated in the study accompanying this package.

Usage

convert_settings_tf_prediction(settings, all_tfs, single = TRUE)

Arguments

Value

A list with following elements: $name, $tf: name of active tf(s) (only if validation is TRUE), $from (tf(s) that will be tested for activity prediction), $response

Examples

## Not run: 
settings <- lapply(expression_settings_validation[1:5], convert_expression_settings_evaluation)
settings_tf_pred <- convert_settings_tf_prediction(settings, all_tfs = c("SMAD1", "STAT1", "RELA"), single = TRUE)
# show how this function can be used to predict activities of TFs
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
tf_target <- construct_tf_target_matrix(weighted_networks, tfs_as_cols = TRUE, standalone_output = TRUE)
tf_importances <- dplyr::bind_rows(lapply(settings_tf_pred, get_single_ligand_importances, tf_target, known = FALSE))
print(head(tf_importances))

## End(Not run)

Description

convert_expression_settings_evaluation Converts expression settings to format in which the total number of potential ligands is reduced up to n top-predicted active ligands.(useful for applications when a lot of ligands are potentially active, a lot of settings need to be predicted and a multi-ligand model is trained).

Usage

convert_settings_topn_ligand_prediction(setting, importances, model, n, normalization)

Arguments

Value

A list with following elements: $name, $from, $response. $response will be a gene-named logical vector indicating whether the gene's transcription was influenced by the active ligand(s) in the setting of interest.

Examples

## Not run: 
settings <- lapply(expression_settings_validation[1:5], convert_expression_settings_evaluation)
settings_ligand_pred <- convert_settings_ligand_prediction(settings, all_ligands = unlist(extract_ligands_from_settings(settings, combination = FALSE)), validation = TRUE, single = TRUE)

weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
ligands <- extract_ligands_from_settings(settings_ligand_pred, combination = FALSE)
ligand_target_matrix <- construct_ligand_target_matrix(weighted_networks, lr_network, ligands)
ligand_importances <- dplyr::bind_rows(lapply(settings_ligand_pred, get_single_ligand_importances, ligand_target_matrix))
evaluation <- evaluate_importances_ligand_prediction(ligand_importances, "median", "lda")

settings <- lapply(expression_settings_validation[5:10], convert_expression_settings_evaluation)
settings_ligand_pred <- convert_settings_ligand_prediction(settings, all_ligands = unlist(extract_ligands_from_settings(settings, combination = FALSE)), validation = FALSE, single = TRUE)
ligands <- extract_ligands_from_settings(settings_ligand_pred, combination = FALSE)
ligand_target_matrix <- construct_ligand_target_matrix(weighted_networks, lr_network, ligands)
ligand_importances <- dplyr::bind_rows(lapply(settings_ligand_pred, get_single_ligand_importances, ligand_target_matrix, known = FALSE))
settings <- lapply(settings, convert_settings_topn_ligand_prediction, importances = ligand_importances, model = evaluation$model, n = 3, normalization = "median")

## End(Not run)

Description

convert_single_cell_expression_to_settings Prepare single-cell expression data to perform ligand activity analysis

Usage

convert_single_cell_expression_to_settings(cell_id, expression_matrix, setting_name, setting_from, regression = FALSE)

Arguments

Value

A list with slots $name, $from and $response respectively containing the setting name, potentially active ligands and the response to predict (whether genes belong to gene set of interest; i.e. most strongly expressed genes in a cell)

Examples

## Not run: 
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
ligands <- list("TNF", "BMP2", "IL4")
ligand_target_matrix <- construct_ligand_target_matrix(weighted_networks, lr_network, ligands, ltf_cutoff = 0, algorithm = "PPR", damping_factor = 0.5, secondary_targets = FALSE)
potential_ligands <- c("TNF", "BMP2", "IL4")
genes <- c("SOCS2", "SOCS3", "IRF1", "ICAM1", "ID1", "ID2", "ID3")
cell_ids <- c("cell1", "cell2")
expression_scaled <- matrix(rnorm(length(genes) * 2, sd = 0.5, mean = 0.5), nrow = 2)
rownames(expression_scaled) <- cell_ids
colnames(expression_scaled) <- genes
settings <- convert_single_cell_expression_to_settings(cell_id = cell_ids[1], expression_matrix = expression_scaled, setting_name = "test", setting_from = potential_ligands)

## End(Not run)

Description

correct_topology_ppr The ligand-target probability scores of a matrix constructed via personalized pagerank will be subtracted by target probability scores calculated via global pagerank; these latter scores can be considered as scores solely attributed to network topology and not by proximity to the ligand of interest. Recommended to use this function in combination with a ligand-target matrix constructed without applying a cutoff on the ligand-tf matrix.

Usage

correct_topology_ppr(ligand_target_matrix,weighted_networks,ligands_position = "cols")

Arguments

Value

A matrix containing ligand-target probability scores, after subtracting target scores solely due to network topology.

Examples

## Not run: 
## Generate the ligand-target matrix from loaded weighted_networks
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
ligands <- list("TNF", "BMP2", c("IL4", "IL13"))
ligand_target_matrix <- construct_ligand_target_matrix(weighted_networks, lr_network, ligands, ltf_cutoff = 0, algorithm = "PPR", damping_factor = 0.5, secondary_targets = FALSE)
ligand_target_matrix <- correct_topology_ppr(ligand_target_matrix, weighted_networks, ligands_position = "cols")

## End(Not run)

Description

diagrammer_format_signaling_graph Prepare extracted ligand-target signaling network for visualization with DiagrammeR.

Usage

diagrammer_format_signaling_graph(signaling_graph_list, ligands_all,targets_all, sig_color = "steelblue", gr_color = "orange")

Arguments

Value

A DiagrammeR Graph object ready for visualization via DiagrammeR::render_graph.

Examples

## Not run: 
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
ligands <- list("TNF", "BMP2", c("IL4", "IL13"))
ligand_tf_matrix <- construct_ligand_tf_matrix(weighted_networks, lr_network, ligands, ltf_cutoff = 0.99, algorithm = "PPR", damping_factor = 0.5, ligands_as_cols = TRUE)
all_ligands <- c("BMP2")
all_targets <- c("HEY1")
top_n_regulators <- 2
ligand_target_signaling_list <- get_ligand_signaling_path(ligand_tf_matrix, all_ligands, all_targets, top_n_regulators, weighted_networks)
graph <- diagrammer_format_signaling_graph(ligand_target_signaling_list, all_ligands, all_targets)
# DiagrammeR::render_graph(graph, layout = "tree")

## End(Not run)

Description

estimate_source_weights_characterization will estimate data source weights of data sources of interest based on a model that was trained to predict weights of data sources based on leave-one-in and leave-one-out characterization performances.

Usage

estimate_source_weights_characterization(loi_performances,loo_performances,source_weights_df, sources_oi, random_forest =FALSE)

Arguments

Value

A list containing two elements. $source_weights_df (the input source_weights_df extended by the estimated source_weighs for data sources of interest) and $model (model object of the regression between leave-one-in, leave-one-out performances and data source weights).

Examples

## Not run: 
library(dplyr)
settings <- lapply(expression_settings_validation[1:4], convert_expression_settings_evaluation)
weights_settings_loi <- prepare_settings_leave_one_in_characterization(lr_network = lr_network, sig_network = sig_network, gr_network = gr_network, source_weights_df)
weights_settings_loi <- lapply(weights_settings_loi, add_hyperparameters_parameter_settings, lr_sig_hub = 0.25, gr_hub = 0.5, ltf_cutoff = 0, algorithm = "PPR", damping_factor = 0.2, correct_topology = TRUE)
doMC::registerDoMC(cores = 4)
job_characterization_loi <- parallel::mclapply(weights_settings_loi[1:4], evaluate_model, lr_network = lr_network, sig_network = sig_network, gr_network = gr_network, settings, calculate_popularity_bias_target_prediction = FALSE, calculate_popularity_bias_ligand_prediction = FALSE, ncitations, mc.cores = 4)
loi_performances <- process_characterization_target_prediction_average(job_characterization_loi)
weights_settings_loo <- prepare_settings_leave_one_out_characterization(lr_network = lr_network, sig_network = sig_network, gr_network = gr_network, source_weights_df)
weights_settings_loo <- lapply(weights_settings_loo, add_hyperparameters_parameter_settings, lr_sig_hub = 0.25, gr_hub = 0.5, ltf_cutoff = 0, algorithm = "PPR", damping_factor = 0.2, correct_topology = TRUE)
doMC::registerDoMC(cores = 4)
job_characterization_loo <- parallel::mclapply(weights_settings_loo[1:4], evaluate_model, lr_network = lr_network, sig_network = sig_network, gr_network = gr_network, settings, calculate_popularity_bias_target_prediction = FALSE, calculate_popularity_bias_ligand_prediction = FALSE, ncitations, mc.cores = 4)
loo_performances <- process_characterization_target_prediction_average(job_characterization_loo)
sources_oi <- c("kegg_cytokines")
output <- estimate_source_weights_characterization(loi_performances, loo_performances, source_weights_df %>% filter(source != "kegg_cytokines"), sources_oi, random_forest = FALSE)

## End(Not run)

Description

evaluate_importances_ligand_prediction Evaluate how well a trained model of ligand importance scores is able to predict the true activity state of a ligand. For this it is assumed, that ligand importance measures for truely active ligands will be higher than for non-active ligands. A classificiation algorithm chosen by the user is trained to construct one model based on the ligand importance scores of all ligands of interest (ligands importance scores are considered as features). Several classification evaluation metrics for the prediction are calculated and variable importance scores can be extracted to rank the different importance measures in order of importance for ligand activity state prediction.

Usage

evaluate_importances_ligand_prediction(importances, normalization, algorithm, var_imps = TRUE, cv = TRUE, cv_number = 4, cv_repeats = 2, parallel = FALSE, n_cores = 4,ignore_errors = FALSE)

Arguments

Value

A list with the following elements. $performances: data frame containing classification evaluation measure for classification on the test folds during training via cross-validation; $performances_training: data frame containing classification evaluation measures for classification of the final model (discrete class assignments) on the complete data set (performance can be severly optimistic due to overfitting!); $performance_training_continuous: data frame containing classification evaluation measures for classification of the final model (class probability scores) on the complete data set (performance can be severly optimistic due to overfitting!) $var_imps: data frame containing the variable importances of the different ligands (embbed importance score for some classification algorithms, otherwise just the auroc); $prediction_response_df: data frame containing for each ligand-setting combination the ligand importance scores for the individual importance scores, the complete model of importance scores and the ligand activity as well (TRUE or FALSE); $model: the caret model object that can be used on new importance scores to predict the ligand activity state.

Examples

## Not run: 
settings <- lapply(expression_settings_validation[1:5], convert_expression_settings_evaluation)
settings_ligand_pred <- convert_settings_ligand_prediction(settings, all_ligands = unlist(extract_ligands_from_settings(settings, combination = FALSE)), validation = TRUE, single = TRUE)

weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
ligands <- extract_ligands_from_settings(settings_ligand_pred, combination = FALSE)
ligand_target_matrix <- construct_ligand_target_matrix(weighted_networks, lr_network, ligands)
ligand_importances <- dplyr::bind_rows(lapply(settings_ligand_pred, get_single_ligand_importances, ligand_target_matrix))
evaluation <- evaluate_importances_ligand_prediction(ligand_importances, "median", "lda")
print(head(evaluation))

## End(Not run)

Description

evaluate_ligand_prediction_per_bin: Evaluate ligand activity predictions for different bins/groups of targets genes. Bins are constructed such that genes that are similarly frequently cited are grouped together and the different bins have similar size.

Usage

evaluate_ligand_prediction_per_bin(nbins,settings,ligand_target_matrix,ncitations,ligands_position = "cols",...)

Arguments

Value

A data.frame containing several classification evaluation metrics for ligand activity prediction. Predictions were evaluated for n different bins of target genes. The specific bin is indicated in the variable target_bin_id. target_bin_id = 1: target genes that are least mentioned in the Pubmed literature.

Examples

## Not run: 
library(dplyr)
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
settings <- lapply(expression_settings_validation[1:10], convert_expression_settings_evaluation)
ligands <- extract_ligands_from_settings(settings)
ligand_target_matrix <- construct_ligand_target_matrix(weighted_networks, lr_network, ligands)
# ncitations = get_ncitations_genes()
ligand_prediction_performances_target_bins_popularity <- evaluate_ligand_prediction_per_bin(5, settings, ligand_target_matrix, ncitations)

## End(Not run)

Description

evaluate_model will take as input a setting of parameters (data source weights and hyperparameters) and layer-specific networks to construct a ligand-target matrix and evaluate its performance on input validation settings (both target gene prediction and ligand activity prediction).

Usage

evaluate_model(parameters_setting, lr_network, sig_network, gr_network, settings, calculate_popularity_bias_target_prediction,calculate_popularity_bias_ligand_prediction, ncitations = ncitations, secondary_targets = FALSE, remove_direct_links = "no", n_target_bins = 3,...)

Arguments

Value

A list containing following elements: $model_name, $performances_target_prediction, $performances_ligand_prediction, $performances_ligand_prediction_single

Examples

## Not run: 
library(dplyr)
settings <- lapply(expression_settings_validation[1:4], convert_expression_settings_evaluation)
weights_settings_loi <- prepare_settings_leave_one_in_characterization(lr_network, sig_network, gr_network, source_weights_df)
weights_settings_loi <- lapply(weights_settings_loi, add_hyperparameters_parameter_settings, lr_sig_hub = 0.25, gr_hub = 0.5, ltf_cutoff = 0, algorithm = "PPR", damping_factor = 0.8, correct_topology = TRUE)
doMC::registerDoMC(cores = 8)
output_characterization <- parallel::mclapply(weights_settings_loi[1:3], evaluate_model, lr_network, sig_network, gr_network, settings, calculate_popularity_bias_target_prediction = TRUE, calculate_popularity_bias_ligand_prediction = TRUE, ncitations, mc.cores = 3)

## End(Not run)

Description

evaluate_model_application will take as input a setting of parameters (data source weights and hyperparameters) and layer-specific networks to construct a ligand-target matrix and evaluate its performance on input application settings (only target gene prediction).

Usage

evaluate_model_application(parameters_setting, lr_network, sig_network, gr_network, settings, secondary_targets = FALSE, remove_direct_links = "no", ...)

Arguments

Value

A list containing following elements: $model_name, $performances_target_prediction.

Examples

## Not run: 
library(dplyr)
settings <- lapply(expression_settings_validation[1:4], convert_expression_settings_evaluation)
weights_settings_loi <- prepare_settings_leave_one_in_characterization(lr_network, sig_network, gr_network, source_weights_df)
weights_settings_loi <- lapply(weights_settings_loi, add_hyperparameters_parameter_settings, lr_sig_hub = 0.25, gr_hub = 0.5, ltf_cutoff = 0, algorithm = "PPR", damping_factor = 0.8, correct_topology = TRUE)
doMC::registerDoMC(cores = 8)
output_characterization <- parallel::mclapply(weights_settings_loi[1:3], evaluate_model_application, lr_network, sig_network, gr_network, settings, mc.cores = 3)

## End(Not run)

Description

evaluate_model_application_multi_ligand will take as input a setting of parameters (data source weights and hyperparameters) and layer-specific networks to construct a ligand-target matrix and evaluate its performance on input application settings (only target gene prediction; multi-ligand classification).

Usage

evaluate_model_application_multi_ligand(parameters_setting, lr_network, sig_network, gr_network, settings, secondary_targets = FALSE, remove_direct_links = "no",classification_algorithm = "lda", ...)

Arguments

Value

A list containing following elements: $model_name, $performances_target_prediction.

Examples

## Not run: 
library(dplyr)
settings <- convert_expression_settings_evaluation(expression_settings_validation$TGFB_IL6_timeseries) %>% list()
weights_settings_loi <- prepare_settings_leave_one_in_characterization(lr_network, sig_network, gr_network, source_weights_df)
weights_settings_loi <- lapply(weights_settings_loi, add_hyperparameters_parameter_settings, lr_sig_hub = 0.25, gr_hub = 0.5, ltf_cutoff = 0, algorithm = "PPR", damping_factor = 0.8, correct_topology = TRUE)
doMC::registerDoMC(cores = 8)
output_characterization <- parallel::mclapply(weights_settings_loi[1:3], evaluate_model_application_multi_ligand, lr_network, sig_network, gr_network, settings, classification_algorithm = "lda", var_imps = FALSE, cv_number = 5, cv_repeats = 4, parallel = TRUE, mc.cores = 3)

## End(Not run)

Description

evaluate_model_cv will take as input a setting of parameters (data source weights and hyperparameters) and layer-specific networks to construct a ligand-target matrix and calculate the model's performance in target gene prediction and feature importance scores for ligand prediction).

Usage

evaluate_model_cv(parameters_setting, lr_network, sig_network, gr_network, settings, secondary_targets = FALSE, remove_direct_links = "no",...)

Arguments

Value

A list containing following elements: $performances_target_prediction, $importances_ligand_prediction.

Examples

## Not run: 
library(dplyr)
settings <- lapply(expression_settings_validation[1:4], convert_expression_settings_evaluation)
weights_settings_loi <- prepare_settings_leave_one_in_characterization(lr_network, sig_network, gr_network, source_weights_df)
weights_settings_loi <- lapply(weights_settings_loi, add_hyperparameters_parameter_settings, lr_sig_hub = 0.25, gr_hub = 0.5, ltf_cutoff = 0, algorithm = "PPR", damping_factor = 0.8, correct_topology = TRUE)
doMC::registerDoMC(cores = 8)
output_characterization <- parallel::mclapply(weights_settings_loi[1:3], evaluate_model_cv, lr_network, sig_network, gr_network, settings, calculate_popularity_bias_target_prediction = TRUE, calculate_popularity_bias_ligand_prediction = TRUE, ncitations, mc.cores = 3)

## End(Not run)

Description

evaluate_multi_ligand_target_prediction Evaluate how well a trained model is able to predict the observed response to a combination of ligands (e.g. the set of DE genes after treatment of cells by multiple ligands). A classificiation algorithm chosen by the user is trained to construct one model based on the target gene predictions of all ligands of interest (ligands are considered as features). Several classification evaluation metrics for the prediction are calculated depending on whether the input ligand-target matrix contains probability scores for targets or discrete target assignments. In addition, variable importance scores can be extracted to rank the possible active ligands in order of importance for response prediction.

Usage

evaluate_multi_ligand_target_prediction(setting,ligand_target_matrix, ligands_position = "cols", algorithm, var_imps = TRUE, cv = TRUE, cv_number = 4, cv_repeats = 2, parallel = FALSE, n_cores = 4,ignore_errors = FALSE,continuous = TRUE)

Arguments

Value

A list with the following elements. $performances: data frame containing classification evaluation measure for classification on the test folds during training via cross-validation; $performances_training: data frame containing classification evaluation measures for classification of the final model (discrete class assignments) on the complete data set (performance can be severly optimistic due to overfitting!); $performance_training_continuous: data frame containing classification evaluation measures for classification of the final model (class probability scores) on the complete data set (performance can be severly optimistic due to overfitting!) $var_imps: data frame containing the variable importances of the different ligands (embbed importance score for some classification algorithms, otherwise just the auroc); $prediction_response_df: data frame containing for each gene the ligand-target predictions of the individual ligands, the complete model and the response as well; $setting: name of the specific setting that needed to be evaluated; $ligands: ligands of interest.

Examples

## Not run: 
library(dplyr)
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
setting <- convert_expression_settings_evaluation(expression_settings_validation$TGFB_IL6_timeseries) %>% list()
ligands <- extract_ligands_from_settings(setting)
ligand_target_matrix <- construct_ligand_target_matrix(weighted_networks, lr_network, ligands)
output <- lapply(setting, evaluate_multi_ligand_target_prediction, ligand_target_matrix, ligands_position = "cols", algorithm = "glm")
output <- lapply(setting, evaluate_multi_ligand_target_prediction, make_discrete_ligand_target_matrix(ligand_target_matrix), ligands_position = "cols", algorithm = "glm")

## End(Not run)

Description

evaluate_multi_ligand_target_prediction_regression Evaluate how well a trained model is able to predict the observed response to a combination of ligands (e.g. the absolute log fold change value of genes after treatment of cells by a ligand). A regression algorithm chosen by the user is trained to construct one model based on the target gene predictions of all ligands of interest (ligands are considered as features). It shows several regression model fit metrics for the prediction. In addition, variable importance scores can be extracted to rank the possible active ligands in order of importance for response prediction.

Usage

evaluate_multi_ligand_target_prediction_regression(setting,ligand_target_matrix, ligands_position = "cols", algorithm, var_imps = TRUE, cv = TRUE, cv_number = 4, cv_repeats = 2, parallel = FALSE, n_cores = 4,ignore_errors = FALSE)

Arguments

Value

A list with the following elements. $performances: data frame containing regression model fit metrics for regression on the test folds during training via cross-validation; $performances_training: data frame containing model fit metrics for regression of the final model on the complete data set (performance can be severly optimistic due to overfitting!); $var_imps: data frame containing the variable importances of the different ligands (embbed importance score for some classification algorithms, otherwise just the auroc); $prediction_response_df: data frame containing for each gene the ligand-target predictions of the individual ligands, the complete model and the response as well; $setting: name of the specific setting that needed to be evaluated; $ligands: ligands of interest.

Examples

## Not run: 
library(dplyr)
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
setting <- convert_expression_settings_evaluation_regression(expression_settings_validation$TGFB_IL6_timeseries) %>% list()
ligands <- extract_ligands_from_settings(setting)
ligand_target_matrix <- construct_ligand_target_matrix(weighted_networks, lr_network, ligands)
output <- lapply(setting, evaluate_multi_ligand_target_prediction_regression, ligand_target_matrix, ligands_position = "cols", algorithm = "lm")

## End(Not run)

Description

evaluate_random_model will take as input a setting of parameters (data source weights and hyperparameters) and layer-specific networks to construct a ligand-target matrix and evaluate its performance on input validation settings (both target gene prediction and ligand activity prediction) after randomisation of the networks by edge swapping.

Usage

evaluate_random_model(parameters_setting, lr_network, sig_network, gr_network, settings, calculate_popularity_bias_target_prediction,calculate_popularity_bias_ligand_prediction, ncitations = ncitations, secondary_targets = FALSE, remove_direct_links = "no", n_target_bins = 3,...)

Arguments

Value

A list containing following elements: $model_name, $performances_target_prediction, $performances_ligand_prediction, $performances_ligand_prediction_single

Examples

## Not run: 
library(dplyr)
settings <- lapply(expression_settings_validation[1:4], convert_expression_settings_evaluation)
weights_settings_loi <- prepare_settings_leave_one_in_characterization(lr_network, sig_network, gr_network, source_weights_df)
weights_settings_loi <- lapply(weights_settings_loi, add_hyperparameters_parameter_settings, lr_sig_hub = 0.25, gr_hub = 0.5, ltf_cutoff = 0, algorithm = "PPR", damping_factor = 0.8, correct_topology = TRUE)
doMC::registerDoMC(cores = 8)
output_characterization <- parallel::mclapply(weights_settings_loi[1:3], evaluate_random_model, lr_network, sig_network, gr_network, settings, calculate_popularity_bias_target_prediction = TRUE, calculate_popularity_bias_ligand_prediction = TRUE, ncitations, mc.cores = 3)

## End(Not run)

Description

evaluate_single_importances_ligand_prediction Evaluate how well a single ligand importance score is able to predict the true activity state of a ligand. For this it is assumed, that ligand importance measures for truely active ligands will be higher than for non-active ligands. Several classification evaluation metrics for the prediction are calculated and variable importance scores can be extracted to rank the different importance measures in order of importance for ligand activity state prediction.

Usage

evaluate_single_importances_ligand_prediction(importances,normalization)

Arguments

Value

A data frame containing classification evaluation measures for the ligand activity state prediction single, individual feature importance measures.

Examples

## Not run: 
settings <- lapply(expression_settings_validation[1:5], convert_expression_settings_evaluation)
settings_ligand_pred <- convert_settings_ligand_prediction(settings, all_ligands = unlist(extract_ligands_from_settings(settings, combination = FALSE)), validation = TRUE, single = TRUE)

weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
ligands <- extract_ligands_from_settings(settings_ligand_pred, combination = FALSE)
ligand_target_matrix <- construct_ligand_target_matrix(weighted_networks, lr_network, ligands)
ligand_importances <- dplyr::bind_rows(lapply(settings_ligand_pred, get_single_ligand_importances, ligand_target_matrix))
evaluation <- evaluate_single_importances_ligand_prediction(ligand_importances, normalization = "median")
print(head(evaluation))

## End(Not run)

Description

evaluate_target_prediction Evaluate how well the model (i.e. the inferred ligand-target probability scores) is able to predict the observed response to a ligand (e.g. the set of DE genes after treatment of cells by a ligand). It shows several classification evaluation metrics for the prediction. Different classification metrics are calculated depending on whether the input ligand-target matrix contains probability scores for targets or discrete target assignments.

Usage

evaluate_target_prediction(setting,ligand_target_matrix, ligands_position = "cols")

Arguments

Value

A data.frame with following variables: setting, ligand nd for probabilistic predictions: auroc, aupr, aupr_corrected (aupr - aupr for random prediction), sensitivity_roc (proxy measure, inferred from ROC), specificity_roc (proxy measure, inferred from ROC), mean_rank_GST_log_pval (-log10 of p-value of mean-rank gene set test), pearson (correlation coefficient), spearman (correlation coefficient); whereas for categorical predictions: accuracy, recall, specificity, precision, F1, F0.5, F2, mcc, informedness, markedness, fisher_pval_log (which is -log10 of p-value fisher exact test), fisher odds.
"mean_rank_GST_log_pval" will only be included in the dataframe if limma is installed. From NicheNet v2.1.7 onwards, limma is no longer a hard dependency of NicheNet.

Examples

## Not run: 
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
setting <- lapply(expression_settings_validation[1], convert_expression_settings_evaluation)
ligands <- extract_ligands_from_settings(setting)
ligand_target_matrix <- construct_ligand_target_matrix(weighted_networks, lr_network, ligands)
perf1 <- lapply(setting, evaluate_target_prediction, ligand_target_matrix)
print(head(perf1))
perf2 <- lapply(setting, evaluate_target_prediction, make_discrete_ligand_target_matrix(ligand_target_matrix))

## End(Not run)

Description

evaluate_target_prediction_interprete Evaluate how well the model (i.e. the inferred ligand-target probability scores) is able to predict the observed response to a ligand (e.g. the set of DE genes after treatment of cells by a ligand; or the log fold change values). It shows several classification evaluation metrics for the prediction when response is categorical, or several regression model fit metrics when the response is continuous.

Usage

evaluate_target_prediction_interprete(setting,ligand_target_matrix, ligands_position = "cols")

Arguments

Value

A list with the elements $performances and $prediction_response_df. $performance is a data.frame with classification evaluation metrics if response is categorical, or regression model fit metrics if response is continuous. $prediction_response_df shows for each gene, the model prediction and the response value of the gene (e.g. whether the gene the gene is a target or not according to the observed response, or the absolute value of the log fold change of a gene).

Examples

## Not run: 
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
setting <- lapply(expression_settings_validation[1], convert_expression_settings_evaluation)
ligands <- extract_ligands_from_settings(setting)
ligand_target_matrix <- construct_ligand_target_matrix(weighted_networks, lr_network, ligands)
perf1 <- lapply(setting, evaluate_target_prediction_interprete, ligand_target_matrix)
setting <- lapply(expression_settings_validation[1], convert_expression_settings_evaluation_regression)
perf2 <- lapply(setting, evaluate_target_prediction_interprete, ligand_target_matrix)

## End(Not run)

Description

evaluate_target_prediction_per_bin: Evaluate target gene predictions for different bins/groups of targets genes. Bins are constructed such that genes that are similarly frequently cited are grouped together and the different bins have similar size.

Usage

evaluate_target_prediction_per_bin(nbins,settings,ligand_target_matrix,ncitations,ligands_position = "cols")

Arguments

Value

A data.frame containing several classification evaluation metrics for target gene prediction. Predictions were evaluated for n different bins of target genes. The specific bin is indicated in the variable target_bin_id. target_bin_id = 1: target genes that are least mentioned in the Pubmed literature.

Examples

## Not run: 
library(dplyr)
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
settings <- lapply(expression_settings_validation[1:10], convert_expression_settings_evaluation)
ligands <- extract_ligands_from_settings(settings)
ligand_target_matrix <- construct_ligand_target_matrix(weighted_networks, lr_network, ligands)
# ncitations = get_ncitations_genes()
performances_target_bins_popularity <- evaluate_target_prediction_per_bin(5, settings, ligand_target_matrix, ncitations)

## End(Not run)

Description

evaluate_target_prediction_regression Evaluate how well the model (i.e. the inferred ligand-target probability scores) is able to predict the observed response to a ligand (e.g. the absolute log fold change value of genes after treatment of cells by a ligand). It shows several regression model fit metrics for the prediction.

Usage

evaluate_target_prediction_regression(setting,ligand_target_matrix, ligands_position = "cols")

Arguments

Value

A data.frame with following variables: setting, ligand and as regression model fit metrics: r_squared: R squared, adj_r_squared: adjusted R squared, f_statistic: estimate of F-statistic, lm_coefficient_abs_t: absolute value of t-value of coefficient, inverse_rse: 1 divided by estimated standard deviation of the errors (inversed to become that higher values indicate better fit), reverse_aic: reverse value of Akaike information criterion (-AIC, to become that higher values indicate better fit), reverse_bic: the reverse value of the bayesian information criterion, inverse_mae: mean absolute error, pearson: pearson correlation coefficient, spearman: spearman correlation coefficient.

Examples

## Not run: 
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
settings <- lapply(expression_settings_validation[1:2], convert_expression_settings_evaluation_regression)
ligands <- extract_ligands_from_settings(settings)
ligand_target_matrix <- construct_ligand_target_matrix(weighted_networks, lr_network, ligands)
perf1 <- lapply(settings, evaluate_target_prediction_regression, ligand_target_matrix)

## End(Not run)

Description

A list

Usage

expression_settings_validation

Format

A list

diffexp

lfc

Log fold change (treated vs untreated)

qval

Fdr corrected p-value

gene

Gene symbol

from

Gene symbol(s) of ligand(s) by which the cells were treated

name

Name of the expression validation dataset

type

Type of dataset based on response time: "primary", "secondary", "primary + secondary"

Description

extract_ligands_from_settings Extract ligands of interest from (expression) settings in correct to construct the ligand-target matrix.

Usage

extract_ligands_from_settings(settings,combination = TRUE)

Arguments

Value

A list containing the ligands and ligands combinations for which a ligand-target matrix should be constructed. When for a particular dataset multiple ligands are possibly active (i.e. more than ligand in .$from slot of sublist of settings), then both the combination of these multiple ligands and each of these multiple ligands individually will be select for model construction.

Examples

## Not run: 
ligands <- extract_ligands_from_settings(expression_settings_validation)

## End(Not run)

Description

extract_top_fraction_ligands Get the predicted top n percentage ligands of a target of interest

Usage

extract_top_fraction_ligands(target_oi,top_fraction,ligand_target_matrix,ligands_position = "cols")

Arguments

Value

A named numeric vector of ligand-target gene probability scores of the top ligands.

Examples

## Not run: 
## Generate the ligand-target matrix from loaded weighted_networks
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
ligands <- list("TNF", "BMP2", c("IL4", "IL13"))
ligand_target_matrix <- construct_ligand_target_matrix(weighted_networks, lr_network, ligands)
targets <- extract_top_fraction_ligands("ID3", 0.01, ligand_target_matrix)

## End(Not run)

Description

extract_top_fraction_targets Get the predicted top n percentage target genes of a ligand of interest.

Usage

extract_top_fraction_targets(ligand_oi,top_fraction,ligand_target_matrix,ligands_position = "cols")

Arguments

Value

A named numeric vector of ligand-target gene probability scores of the top target genes.

Examples

## Not run: 
## Generate the ligand-target matrix from loaded weighted_networks
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
ligands <- list("TNF", "BMP2", c("IL4", "IL13"))
ligand_target_matrix <- construct_ligand_target_matrix(weighted_networks, lr_network, ligands)
targets <- extract_top_fraction_targets("BMP2", 0.01, ligand_target_matrix)

## End(Not run)

Description

extract_top_n_ligands Get the predicted top n ligands of a target gene of interest.

Usage

extract_top_n_ligands(target_oi,top_n,ligand_target_matrix,ligands_position = "cols")

Arguments

Value

A named numeric vector of ligand-target gene probability scores of the top ligands.

Examples

## Not run: 
## Generate the ligand-target matrix from loaded weighted_networks
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
ligands <- list("TNF", "BMP2", c("IL4", "IL13"))
ligand_target_matrix <- construct_ligand_target_matrix(weighted_networks, lr_network, ligands)
targets <- extract_top_n_ligands("BMP2", 2, ligand_target_matrix)

## End(Not run)

Description

extract_top_n_targets Get the predicted top n target genes of a ligand of interest.

Usage

extract_top_n_targets(ligand_oi,top_n,ligand_target_matrix,ligands_position = "cols")

Arguments

Value

A named numeric vector of ligand-target gene probability scores of the top target genes.

Examples

## Not run: 
## Generate the ligand-target matrix from loaded weighted_networks
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
ligands <- list("TNF", "BMP2", c("IL4", "IL13"))
ligand_target_matrix <- construct_ligand_target_matrix(weighted_networks, lr_network, ligands)
targets <- extract_top_n_targets("BMP2", 50, ligand_target_matrix)

## End(Not run)

Description

A data.frame/tibble describing HGNC human gene symbols, their entrez ids and the MGI mouse gene symbols and entrez ids of the one-one orthologs as determined via NCBI's homologene db and biomaRt ensembl db.

Usage

geneinfo_2022

Format

A data frame/tibble

symbol

human gene symbol

entrez

human gene entrez

entrez_mouse

mouse homolog gene entrez

symbol_mouse

mouse homolog gene symbol

Description

A data.frame/tibble describing HGNC human gene symbols, their entrez ids and potential aliases.

Usage

geneinfo_alias_human

Format

A data frame/tibble

symbol

human gene symbol

entrez

human gene entrez

alias

human gene alias

Description

A data.frame/tibble describing MGI mouse gene symbols, their entrez ids and potential aliases.

Usage

geneinfo_alias_mouse

Format

A data frame/tibble

symbol

mouse gene symbol

entrez

mouse gene entrez

alias

mouse gene alias

Description

A data.frame/tibble describing HGNC human gene symbols, their entrez ids and the MGI mouse gene symbols and entrez ids of the one-one orthologs as determined via NCBI's homologene db and biomaRt ensembl db.

Usage

geneinfo_human

Format

A data frame/tibble

symbol

human gene symbol

entrez

human gene entrez

entrez_mouse

mouse homolog gene entrez

symbol_mouse

mouse homolog gene symbol

Description

Calculate differential expression, average expression, and condition specificity of ligands and receptors.

Usage

generate_info_tables(
  seuratObj,
  celltype_colname,
  senders_oi,
  receivers_oi,
  lr_network_filtered,
  condition_colname = NULL,
  condition_oi = NULL,
  condition_reference = NULL,
  scenario = "case_control",
  assay_oi = NULL,
  ...
)

Arguments

Value

List of dataframes containing sender-receiver DE, sender-receiver expression, and condition DE

Description

User can choose the importance attached to each of the following prioritization criteria: differential expression of ligand and receptor, cell-type specificity of expression of ligand and receptor, NicheNet ligand activity

Usage

generate_prioritization_tables(
  sender_receiver_info,
  sender_receiver_de,
  ligand_activities,
  lr_condition_de = NULL,
  prioritizing_weights = NULL,
  scenario = "case_control"
)

Arguments

Value

Data frames of prioritized sender-ligand-receiver-receptor interactions. The resulting dataframe contains columns from the input dataframes, but columns from lr_condition_de are suffixed with _group (some columns from lr_condition_de are also not present). Additionally, the following columns are added:

• lfc_pval_*: product of -log10(pval) and the LFC of the ligand/receptor

• p_val_adapted_*: p-value adapted to the sign of the LFC to only consider interactions where the ligand/receptor is upregulated in the sender/receiver

• activity_zscore: z-score of the ligand activity

• prioritization_score: The prioritization score for each interaction, calculated as a weighted sum of the prioritization criteria.

Moreover, scaled_* columns are scaled using the corresponding column's ranking or the scale_quantile_adapted function. The columns used for prioritization are scaled_p_val_adapted_ligand, scaled_p_val_adapted_receptor, scaled_activity, scaled_avg_exprs_ligand, scaled_avg_exprs_receptor, scaled_p_val_adapted_ligand_group, scaled_p_val_adapted_receptor_group

Examples

## Not run: 

# Calculate tables with generate_info_tables
info_tables <- generate_info_tables(seurat_obj, "celltype", sender_celltypes, receiver, expressed_ligands, expressed_receptors, lr_network, "aggregate", "LCMV", "SS")

# Generate prioritization tables
generate_prioritization_tables(info_tables$sender_receiver_info,
  info_tables$sender_receiver_de,
  ligand_activities,
  info_tables$lr_condition_de,
  scenario = "case_control"
)

# Alternatively, these tables can also be calculated manually:
# Calculate LCMV-specific average expression
expression_info <- get_exprs_avg(seurat_obj, "celltype", condition_oi = "LCMV", condition_colname = "aggregate")

# Calculate LCMV-specific cell-type markers
DE_table <- calculate_de(seurat_obj, "celltype", condition_oi = "LCMV", condition_colname = "aggregate")

# Calculate condition-specific markers
condition_markers <- FindMarkers(
  object = seuratObj, ident.1 = "LCMV", ident.2 = "SS",
  group.by = "aggregate", min.pct = 0, logfc.threshold = 0
) %>% rownames_to_column("gene")

# Process tables
processed_expr_info <- process_table_to_ic(expression_info, table_type = "expression", lr_network)
processed_DE_table <- process_table_to_ic(DE_table,
  table_type = "celltype_DE", lr_network,
  senders_oi = sender_celltypes, receivers_oi = receiver
)
processed_condition_DE_table <- process_table_to_ic(condition_markers, table_type = "group_DE", lr_network)

# Custom weights
prioritizing_weights <- c("de_ligand" = 1, "de_receptor" = 1, "activity_scaled" = 2, "exprs_ligand" = 1, "exprs_receptor" = 1, "ligand_condition_specificity" = 0.5, "receptor_condition_specificity" = 0.5)

generate_prioritization_tables(processed_expr_info, processed_DE_table, ligand_activities, processed_condition_DE_table, prioritizing_weights)

## End(Not run)

Description

get_active_ligand_receptor_network Get active ligand-receptor network by looking for which ligands are expressed in a sender/signaling cell and which receptors are expressed in the receiver cell. Instead of looking at absolute expression, it is possible as well to extract a ligand-receptor network of differentially expressed ligands and receptors if a vector of log2 fold change values is used as input.

Usage

get_active_ligand_receptor_network(expression_sender, expression_receiver, lr_network, expression_cutoff_sender = 0, expression_cutoff_receiver = 0)

Arguments

Value

A data frame containing at least the variables from, to, sender_expression, receiver_expression. In this network, the active ligand-receptor interactions in the system of interest are shown.

Examples

## Not run: 
library(dplyr)
expression_vector_sender <- rnorm(n = 10000, mean = 6, sd = 3)
expression_vector_receiver <- rnorm(n = 10000, mean = 6, sd = 3)
names(expression_vector_sender) <- sample(x = geneinfo_human$symbol, size = 10000, replace = FALSE)
names(expression_vector_receiver) <- sample(x = geneinfo_human$symbol, size = 10000, replace = FALSE)
weighted_lr_network <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df, n_output_networks = 3) %>% .$lr
sender_cell_receiver_lr_network <- get_active_ligand_receptor_network(expression_vector_sender, expression_vector_receiver, weighted_lr_network, expression_cutoff_sender = 4, expression_cutoff_receiver = 4)

## End(Not run)

Description

get_active_ligand_target_df Get active ligand-target network, meaning that only target genes that are part of the response of interest will be kept as target genes. In addition, target genes with probability scores beneath a predefined cutoff will be removed from the nework.

Usage

get_active_ligand_target_df(response,ligand_target_matrix, ligands_position = "cols", cutoff = 0)

Arguments

Value

A data frame representing the active ligand-target network; with variables $ligand, $target and $score.

Examples

## Not run: 
library(dplyr)
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
setting <- lapply(expression_settings_validation[1:2], convert_expression_settings_evaluation)
ligands <- extract_ligands_from_settings(setting)
ligand_target_matrix <- construct_ligand_target_matrix(weighted_networks, lr_network, ligands)
ligand_target_matrix_discrete <- make_discrete_ligand_target_matrix(ligand_target_matrix)
active_lt_df <- get_active_ligand_target_df(setting[[1]] %>% .$response, ligand_target_matrix_discrete)

## End(Not run)

Description

get_active_ligand_target_matrix Get active ligand-target matrix, meaning that only target genes part of the response of interest will be kept as target genes in the input ligand-target matrix.

Usage

get_active_ligand_target_matrix(response,ligand_target_matrix, ligands_position = "cols")

Arguments

Value

A matrix with ligand-target probability scores (or discrete ligand-target assignments) for the active target genes in the system of interest.

Examples

## Not run: 
library(dplyr)
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
setting <- lapply(expression_settings_validation[1:2], convert_expression_settings_evaluation)
ligands <- extract_ligands_from_settings(setting)
ligand_target_matrix <- construct_ligand_target_matrix(weighted_networks, lr_network, ligands)
active_lt <- get_active_ligand_target_matrix(setting[[1]] %>% .$response, ligand_target_matrix)

## End(Not run)

Description

get_active_regulatory_network Get active regulatory network by looking for which TFs/regulatory genes are expressed in the receiver cell. Instead of looking at absolute expression, it is possible as well to extract a gene regulatory network of differentially expressed TFs if a vector of log2 fold change values is used as input.

Usage

get_active_regulatory_network(expression_receiver, gr_network, expression_cutoff_receiver = 0)

Arguments

Value

A data frame containing at least the variables from, to, receiver_expression. In this network, the active gene regulatory interactions in the system of interest are shown.

Examples

## Not run: 
library(dplyr)
expression_vector_receiver <- rnorm(n = 10000, mean = 6, sd = 3)
names(expression_vector_receiver) <- sample(x = geneinfo_human$symbol, size = 10000, replace = FALSE)
receiver_gr_network <- get_active_regulatory_network(expression_vector_receiver, gr_network, expression_cutoff_receiver = 4)

## End(Not run)

Description

get_active_signaling_network Get active signaling network by looking for which signaling-related genes are expressed in the receiver cell. Instead of looking at absolute expression, it is possible as well to extract a signaling network of differentially expressed signaling mediators if a vector of log2 fold change values is used as input.

Usage

get_active_signaling_network(expression_receiver, sig_network, expression_cutoff_receiver = 0)

Arguments

Value

A data frame containing at least the variables from, to, receiver_expression. In this network, the active signaling interactions in the system of interest are shown.

Examples

## Not run: 
library(dplyr)
expression_vector_receiver <- rnorm(n = 10000, mean = 6, sd = 3)
names(expression_vector_receiver) <- sample(x = geneinfo_human$symbol, size = 10000, replace = FALSE)
receiver_sig_network <- get_active_signaling_network(expression_vector_receiver, sig_network, expression_cutoff_receiver = 4)

## End(Not run)

Description

Shows information about cached databases.

Usage

get_database_cache_stats()

Description

Return the genes that are expressed in given cell cluster(s) based on the fraction of cells in the cluster(s) that should express the cell.

Usage

get_expressed_genes(celltype_oi, object, ...)

## Default S3 method:
get_expressed_genes(celltype_oi, object, celltype_annot, pct = 0.1)

## S3 method for class 'Seurat'
get_expressed_genes(celltype_oi, seurat_obj, pct = 0.1, assay_oi = NULL, ...)

Arguments

Value

A vector containing gene symbols of the expressed genes

Examples

## Not run: 
# For sparse matrix
get_expressed_genes("CD8 T", GetAssayData(seuratObj), seuratObj$celltype, pct = 0.10)

## End(Not run)

## Not run: 
# For Seurat object
get_expressed_genes(celltype_oi = "CD8 T", seurat_obj = seuratObj, pct = 0.10)

## End(Not run)

Description

get_exprs_avg Calculate average of gene expression per cell type. If condition_oi is provided, only consider cells from that condition.

Usage

get_exprs_avg(
  seurat_obj,
  celltype_colname,
  condition_oi = NULL,
  condition_colname = NULL,
  assay_oi = NULL,
  ...
)

Arguments

Value

Data frame with average gene expression per cell type.

Examples

## Not run: 
seurat_obj <- readRDS(url("https://zenodo.org/record/3531889/files/seuratObj.rds"))
seurat_obj$celltype <- make.names(seuratObj$celltype)
# Calculate average expression across conditions
expression_info <- get_exprs_avg(seurat_obj, "celltype")
# Calculate LCMV-specific average expression
expression_info <- get_exprs_avg(seurat_obj, "celltype", condition_oi = "LCMV", condition_colname = "aggregate")

## End(Not run)

Description

get_lfc_celltype Get log fold change of genes between two conditions in cell type of interest when using a Seurat single-cell object.

Usage

get_lfc_celltype(celltype_oi, seurat_obj, condition_colname, condition_oi, condition_reference, celltype_col = "celltype")
#'

Arguments

Value

A tbl with the log fold change values of genes. Positive lfc values: higher in condition_oi compared to condition_reference.

Examples

## Not run: 
requireNamespace("dplyr")
seuratObj <- readRDS(url("https://zenodo.org/record/3531889/files/seuratObj_test.rds"))
get_lfc_celltype(seurat_obj = seuratObj, celltype_oi = "CD8 T", condition_colname = "aggregate", condition_oi = "LCMV", condition_reference = "SS", celltype_col = "celltype", expression_pct = 0.10)

## End(Not run)

Description

get_ligand_activities_targets Calculate the ligand activities and infer ligand-target links based on a list of niche-specific genes per receiver cell type.

Usage

get_ligand_activities_targets(niche_geneset_list, ligand_target_matrix, top_n_target)

Arguments

Value

A tibble of ligands, their activities and targets in each receiver cell type

Examples

## Not run: 
seurat_obj <- readRDS(url("https://zenodo.org/record/5840787/files/seurat_obj_subset_integrated_zonation.rds"))
niches <- list(
  "KC_niche" = list(
    "sender" = c("LSECs_portal", "Hepatocytes_portal", "Stellate cells_portal"),
    "receiver" = c("KCs")
  ),
  "MoMac2_niche" = list(
    "sender" = c("Cholangiocytes", "Fibroblast 2"),
    "receiver" = c("MoMac2")
  ),
  "MoMac1_niche" = list(
    "sender" = c("Capsule fibroblasts", "Mesothelial cells"),
    "receiver" = c("MoMac1")
  )
)
DE_receiver <- calculate_niche_de(seurat_obj, niches, "receiver")
expression_pct <- 0.10
lfc_cutoff <- 0.15 # recommended for 10x as min_lfc cutoff.
specificity_score_targets <- "min_lfc"
DE_receiver_processed_targets <- process_receiver_target_de(DE_receiver_targets = DE_receiver, niches = niches, expression_pct = expression_pct, specificity_score = specificity_score_targets)
background <- DE_receiver_processed_targets %>%
  pull(target) %>%
  unique()
geneset_KC <- DE_receiver_processed_targets %>%
  filter(receiver == niches$KC_niche$receiver & target_score >= lfc_cutoff & target_significant == 1 & target_present == 1) %>%
  pull(target) %>%
  unique()
geneset_MoMac2 <- DE_receiver_processed_targets %>%
  filter(receiver == niches$MoMac2_niche$receiver & target_score >= lfc_cutoff & target_significant == 1 & target_present == 1) %>%
  pull(target) %>%
  unique()
geneset_MoMac1 <- DE_receiver_processed_targets %>%
  filter(receiver == niches$MoMac1_niche$receiver & target_score >= lfc_cutoff & target_significant == 1 & target_present == 1) %>%
  pull(target) %>%
  unique()
top_n_target <- 250
niche_geneset_list <- list(
  "KC_niche" = list(
    "receiver" = "KCs",
    "geneset" = geneset_KC,
    "background" = background
  ),
  "MoMac1_niche" = list(
    "receiver" = "MoMac1",
    "geneset" = geneset_MoMac1,
    "background" = background
  ),
  "MoMac2_niche" = list(
    "receiver" = "MoMac2",
    "geneset" = geneset_MoMac2,
    "background" = background
  )
)
ligand_activities_targets <- get_ligand_activities_targets(niche_geneset_list = niche_geneset_list, ligand_target_matrix = ligand_target_matrix, top_n_target = top_n_target)

## End(Not run)

Description

get_ligand_signaling_path Extract possible signaling paths between a ligand and target gene of interest. The most highly weighted path(s) will be extracted.

Usage

get_ligand_signaling_path(ligand_tf_matrix, ligands_all, targets_all, top_n_regulators = 4, weighted_networks, ligands_position = "cols")

Arguments

Value

A list containing 2 elements (sig and gr): the integrated weighted ligand-signaling and gene regulatory networks data frame / tibble format with columns: from, to, weight

Examples

## Not run: 
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
ligands <- list("TNF", "BMP2", c("IL4", "IL13"))
ligand_tf_matrix <- construct_ligand_tf_matrix(weighted_networks, lr_network, ligands, ltf_cutoff = 0.99, algorithm = "PPR", damping_factor = 0.5, ligands_as_cols = TRUE)
all_ligands <- c("BMP2")
all_targets <- c("HEY1")
top_n_regulators <- 2
ligand_target_signaling_list <- get_ligand_signaling_path(ligand_tf_matrix, all_ligands, all_targets, top_n_regulators, weighted_networks)

## End(Not run)

Description

get_ligand_signaling_path_with_receptor Extract possible signaling paths between a ligand(s), receptor(s) and target gene(s) of interest. The most highly weighted path(s) will be extracted.

Usage

get_ligand_signaling_path_with_receptor(ligand_tf_matrix, ligands_all, receptors_all, targets_all, top_n_regulators = 3, weighted_networks, ligands_position = "cols")

Arguments

Value

A list containing 2 elements (sig and gr): the integrated weighted ligand-signaling and gene regulatory networks data frame / tibble format with columns: from, to, weight

Examples

## Not run: 
weighted_networks <- construct_weighted_networks(lr_network, sig_network, gr_network, source_weights_df)
ligands <- list("TNF", "BMP2", c("IL4", "IL13"))
ligand_tf_matrix <- construct_ligand_tf_matrix(weighted_networks, lr_network, ligands, ltf_cutoff = 0.99, algorithm = "PPR", damping_factor = 0.5, ligands_as_cols = TRUE)
all_ligands <- c("BMP2")
all_receptors <- c("BMPR2")
all_targets <- c("HEY1")
top_n_regulators <- 2
ligand_target_signaling_list <- get_ligand_signaling_path_with_receptor(ligand_tf_matrix, all_ligands, all_receptors, all_targets, top_n_regulators, weighted_networks)

## End(Not run)