Calculate ligand activity performance without considering evaluation datasets belonging to the top i most frequently cited ligands

ligand_activity_performance_top_i_removed: calculates ligand activity performance (given ligand importances measures) without considering evaluation datasets belonging to the top i most frequently cited ligands.

Usage

ligand_activity_performance_top_i_removed(i, importances, ncitations)

Arguments

i

Indicate the number of most popular ligands for which the corresponding datasets will be removed for performance calculation.

importances

Data frame containing the ligand importance measures for every ligand-dataset combination. See get_single_ligand_importances and get_multi_ligand_importances.

ncitations

A data frame denoting the number of times a gene is mentioned in the Pubmed literature. Should at least contain following variables: 'symbol' and 'ncitations'. Default: ncitations (variable contained in this package). See function get_ncitations_genes for a function that makes this data frame from current Pubmed information.

Value

A data.frame in which you can find several evaluation metrics of the ligand activity prediction performance and an indication of what percentage of most popular ligands has not been considered ($popularity_index).

Examples

if (FALSE) { # \dontrun{
library(dplyr)
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))

ligand_activity_popularity_bias <- lapply(0:3, ligand_activity_performance_top_i_removed, ligand_importances, ncitations) %>% bind_rows()
# example plot
# ligand_activity_popularity_bias %>% ggplot(aes(popularity_index,aupr)) + geom_smooth(method = "lm") + geom_point()
} # }