Nothing
#' Merge DrugBank and OnSIDES Database Objects
#'
#' Creates an integrated dvobject object by linking DrugBank dvobject with
#' OnSIDES dvobject using RxNorm CUIs as the bridge.
#'
#' @details
#' This function performs the following key steps:
#' 1. Creates a mapping table between DrugBank IDs and RxNorm CUIs from the DrugBank object.
#' 2. Enriches the relevant OnSIDES tables (`vocab_rxnorm_ingredient` and optionally
#' `high_confidence`) by adding a `drugbank_id` column.
#' 3. Assembles a new list object containing all original tables plus the enriched ones
#' and the ID mapping table itself.
#'
#' The resulting object allows for powerful queries that span both mechanistic data from
#' DrugBank and clinical side-effect data from OnSIDES.
#' Supports piping and chaining with other merge functions.
#'
#' @param db_object A dvobject from `parseDrugBank()` OR an existing merged
#' dvobject (containing `$drugbank`).
#' @param onsides_db A dvobject produced by `dbparser::parseOnSIDES()`.
#'
#' @return A new dvobject containing the integrated data.
#'
#' @export
#' @family mergers
#' @importFrom dplyr filter select rename mutate left_join .data
#'
#' @examples
#' \dontrun{
#' # First, parse the individual databases
#' drugbank <- parseDrugBank("path/to/drugbank.xml")
#' onsides <- parseOnSIDES("path/to/onsides_csvs/")
#'
#' # Now, merge them into a single, powerful object
#' merged_db <- merge_drugbank_onsides(drugbank, onsides)
#'
#' # --- Example Analysis: Find the protein targets of all drugs known to ---
#' # --- cause the side effect "Hepatitis" with high confidence. ---
#'
#' # 1. Find the MedDRA ID for "Hepatitis"
#' hepatitis_id <- merged_db$onsides$vocab_meddra_adverse_effect %>%
#' filter(meddra_name == "Hepatitis") %>%
#' pull(meddra_id)
#'
#' # 2. Find all drug ingredients linked to this effect in the high_confidence table
#' drug_ids_causing_hepatitis <- merged_db$onsides$high_confidence_enriched %>%
#' filter(effect_meddra_id == hepatitis_id) %>%
#' pull(drugbank_id) %>%
#' na.omit() %>%
#' unique()
#'
#' # 3. Look up the targets for these DrugBank IDs
#' targets_of_interest <- merged_db$targets %>%
#' filter(parent_key %in% drug_ids_causing_hepatitis) %>%
#' select(drug_id = parent_key, target_name = name, gene_name)
#'
#' head(targets_of_interest)
#' }
merge_drugbank_onsides <- function(db_object, onsides_db) {
# --- Step 0: Input Validation and Hub Detection ---
# This logic enables the Pipe (%>%) and Chaining.
if ("drugbank" %in% names(db_object)) {
# CASE A: Input is an already-merged object (e.g., passed via pipe from another merge)
drugbank_db <- db_object$drugbank
merged_object <- db_object # Start with existing data to preserve previous merges
} else {
# CASE B: Input is a raw DrugBank object
drugbank_db <- db_object
merged_object <- init_dvobject()
merged_object$drugbank <- db_object
attr(merged_object, "DrugBankDB") <- attr(drugbank_db, "original_db_info")
merged_object <- set_source_metadata(merged_object,
attr(drugbank_db, "source_metadata"))
}
# Validate the Hub
if (!inherits(drugbank_db, "dvobject") ||
(!"drugs" %in% names(drugbank_db))) {
stop("`db_object` must contain a valid DrugBank dvobject.")
}
if (!inherits(drugbank_db, "dvobject") ||
(!"external_identifiers" %in% names(drugbank_db$drugs))) {
stop("`drugbank_db` must contain external_identifiers data.")
}
# Validate the Spoke
if (!inherits(onsides_db, "dvobject") ||
(!"vocab_rxnorm_ingredient" %in% names(onsides_db))) {
stop("`onsides_db` must be a valid dvobject from parseOnSIDES().")
}
# --- Step 1: Create the Bridge (RxCUI Mapping Table) ---
message("Creating DrugBank ID <-> RxCUI mapping table...")
rxcui_mapping_df <- drugbank_db$drugs$external_identifiers %>%
dplyr::filter(.data$resource == "RxCUI") %>%
dplyr::select(all_of("drugbank_id"), rxcui = "identifier") %>%
dplyr::distinct()
# --- Step 2: Enrich OnSIDES Tables ---
message("Enriching OnSIDES tables with DrugBank IDs...")
# Enrich the core ingredient vocabulary
onsides_ingredient_enriched <- onsides_db$vocab_rxnorm_ingredient %>%
dplyr::left_join(rxcui_mapping_df, by = c("rxnorm_id" = "rxcui"))
# Optionally enrich the high_confidence table if it exists
if ("high_confidence" %in% names(onsides_db)) {
onsides_hc_enriched <- onsides_db$high_confidence %>%
dplyr::mutate(ingredient_id = as.character(.data$ingredient_id)) %>%
dplyr::left_join(rxcui_mapping_df, by = c("ingredient_id" = "rxcui"))
}
# --- Step 3: Assemble the Final Merged Object ---
message("Assembling final merged object...")
# Add OnSIDES structure (Initialize if needed, but append to existing)
if (is.null(merged_object$onsides)) {
merged_object$onsides <- list()
}
# Copy all OnSIDES tables
for (name in names(onsides_db)) {
merged_object$onsides[[name]] <- onsides_db[[name]]
}
# Ensure integrated_data list exists
if (is.null(merged_object$integrated_data)) {
merged_object$integrated_data <- list()
}
merged_object$integrated_data[["vocab_rxnorm_ingredient_enriched"]] <- onsides_ingredient_enriched
if (exists("onsides_hc_enriched")) {
merged_object$integrated_data[["high_confidence_enriched"]] <- onsides_hc_enriched
}
# Add the mapping table itself for user reference
merged_object$integrated_data[["DrugBank_RxCUI_Mapping"]] <- rxcui_mapping_df
# Update metadata
attr(merged_object, "onSideDB") <- attr(onsides_db, "original_db_info")
merged_object <- set_source_metadata(merged_object,
attr(onsides_db, "source_metadata"))
# Assign a new class (Prepend to keep existing classes like DrugBankTWOSIDESDb)
class(merged_object) <- unique(c("DrugBankOnSIDESDb", class(merged_object)))
message("Merge complete.")
merged_object
}
#' Merge a DrugBank dvobject with a TWOSIDES dvobject
#'
#' Integrates drug-drug interaction data from TWOSIDES with the rich mechanistic
#' information from DrugBank. This function is chainable and can accept a raw
#' DrugBank object or an already-merged dvobject.
#'
#' @param db_object A dvobject from `parseDrugBank()` or an existing merged dvobject.
#' @param twosides_db A dvobject from `parseTWOSIDES()`.
#'
#' @return A new, nested dvobject with the TWOSIDES data added.
#'
#' @importFrom dplyr filter select rename mutate left_join .data distinct
#' @family mergers
#' @export
merge_drugbank_twosides <- function(db_object, twosides_db) {
# --- Step 0: Input Validation and Hub Detection (Pipe Friendly) ---
if ("drugbank" %in% names(db_object)) {
# Case A: Input is an already-merged object
drugbank_db <- db_object$drugbank
merged_object <- db_object
} else {
# Case B: Input is a raw DrugBank object
drugbank_db <- db_object
merged_object <- init_dvobject()
merged_object$drugbank <- db_object
attr(merged_object, "DrugBankDB") <- attr(drugbank_db, "original_db_info")
merged_object <- set_source_metadata(merged_object,
attr(drugbank_db, "source_metadata"))
}
# Validate inputs
if (!inherits(drugbank_db, "dvobject") || (!"drugs" %in% names(drugbank_db))) {
stop("`db_object` must contain a valid DrugBank dvobject.")
}
if (!inherits(drugbank_db, "dvobject") || (!"external_identifiers" %in% names(drugbank_db$drugs))) {
stop("`drugbank_db` must contain external_identifiers data.")
}
if (!is.list(twosides_db) || !("drug_drug_interactions" %in% names(twosides_db))) {
stop("`twosides_db` must be a valid dvobject from parseTWOSIDES().")
}
# --- Step 1: Create Bridge ---
message("Creating DrugBank ID <-> RxCUI mapping table...")
rxcui_mapping_df <- drugbank_db$drugs$external_identifiers %>%
dplyr::filter(.data$resource == "RxCUI") %>%
dplyr::select(all_of("drugbank_id"), rxcui = "identifier") %>%
dplyr::mutate(rxcui = .data$rxcui) %>%
dplyr::distinct()
# Drug name lookup
drug_name_lookup <- drugbank_db$drugs$general_information %>%
dplyr::select(all_of("drugbank_id"), drug_name = "name")
# --- Step 2: Enrich TWOSIDES Data ---
message("Enriching TWOSIDES data with DrugBank information...")
# Prepare lookup tables for double joining
rxcui_map_1 <- rxcui_mapping_df %>% dplyr::rename(drugbank_id_1 = "drugbank_id")
rxcui_map_2 <- rxcui_mapping_df %>% dplyr::rename(drugbank_id_2 = "drugbank_id")
drug_name_lookup_1 <- drug_name_lookup %>%
dplyr::rename(drug_name_1 = "drug_name", drugbank_id_1 = "drugbank_id")
drug_name_lookup_2 <- drug_name_lookup %>%
dplyr::rename(drug_name_2 = "drug_name", drugbank_id_2 = "drugbank_id")
enriched_ddis <- twosides_db$drug_drug_interactions %>%
dplyr::mutate(drug_1_rxnorn_id = as.character(.data$drug_1_rxnorn_id),
drug_2_rxnorm_id = as.character(.data$drug_2_rxnorm_id)) %>%
# LOGIC CHANGE 1: Union (OR) Filter
# Note the spelling: rxnorn
dplyr::filter((.data$drug_1_rxnorn_id %in% rxcui_mapping_df$rxcui) |
(.data$drug_2_rxnorm_id %in% rxcui_mapping_df$rxcui)) %>%
# Join for Drug 1 (using 'rxnorn' spelling)
dplyr::left_join(rxcui_map_1, by = c("drug_1_rxnorn_id" = "rxcui")) %>%
dplyr::left_join(drug_name_lookup_1, by = "drugbank_id_1") %>%
# Join for Drug 2 (Twosides seems to use 'rxnorm')
dplyr::left_join(rxcui_map_2, by = c("drug_2_rxnorm_id" = "rxcui")) %>%
dplyr::left_join(drug_name_lookup_2, by = "drugbank_id_2") %>%
# LOGIC CHANGE 2: Keep if at least one side matched
dplyr::filter(!is.na(.data$drugbank_id_1) | !is.na(.data$drugbank_id_2)) %>%
# LOGIC CHANGE 3: Fallback names to prevent NAs
dplyr::mutate(
drug_name_1 = dplyr::coalesce(.data$drug_name_1, .data$drug_1_concept_name),
drug_name_2 = dplyr::coalesce(.data$drug_name_2, .data$drug_2_concept_name)
)
# --- Step 3: Assemble Final Object ---
# Initialize integrated_data if it doesn't exist
if (is.null(merged_object$integrated_data)) {
merged_object$integrated_data <- list()
}
# Add the new enriched table
merged_object$integrated_data$drug_drug_interactions <- enriched_ddis
# Add raw Twosides data
merged_object$twosides <- twosides_db
# --- Step 4: Metadata ---
attr(merged_object, "TwoSidesDB") <- attr(twosides_db, "original_db_info")
merged_object <- set_source_metadata(merged_object,
attr(twosides_db, "source_metadata"))
# Prepend new class
class(merged_object) <- unique(c("DrugBankTWOSIDESDb", class(merged_object)))
message("Merge complete.")
merged_object
}
# =============================================================================
# File: R/merge_drugbank_hpo.R
# Purpose: Integrate DrugBank with HPO gene-phenotype associations using
# gene symbols as the bridge identifier.
#
# DrugBank structure navigated:
# $cett$targets$general_information → target_id, drugbank_id
# $cett$targets$polypeptides$
# general_information → target_id, gene_name
#
# Join path:
# polypeptides.gene_name + polypeptides.target_id
# → targets.target_id + targets.drugbank_id
# → HPO.gene_symbol
#
# Author: Mohammed Ali — Interstellar Consultation Services
# License: MIT
# =============================================================================
#' Merge DrugBank and HPO Gene-Phenotype Database Objects
#'
#' Creates an integrated dvobject by linking DrugBank target gene data with
#' HPO gene-phenotype associations using gene symbols as the bridge.
#'
#' @details
#' This function performs the following key steps:
#' 1. Extracts gene names from DrugBank polypeptide data
#' (\code{$cett$targets$polypeptides$general_information}).
#' 2. Links gene names to DrugBank IDs via the target_id bridge
#' (\code{$cett$targets$general_information}).
#' 3. Optionally resolves DrugBank IDs to drug names.
#' 4. Enriches the HPO gene-phenotype association table by adding
#' \code{drugbank_id} and \code{drug_name} columns.
#' 5. Assembles a merged object containing all original data plus the
#' enriched table and the gene mapping bridge.
#'
#' Supports piping and chaining with other merge functions.
#'
#' @param db_object A dvobject from \code{parseDrugBank()} OR an existing
#' merged dvobject (containing \code{$drugbank}).
#' @param hpo_db A dvobject produced by \code{dbparser::parseHPO()}.
#'
#' @return A new dvobject containing the integrated data.
#'
#' @export
#' @family mergers
#' @importFrom dplyr filter select mutate left_join inner_join distinct
#' n_distinct all_of .data %>%
#' @importFrom tibble as_tibble
#'
#' @examples
#' \dontrun{
#' drugbank <- parseDrugBank("path/to/drugbank.xml")
#' hpo <- parseHPO("path/to/gene_attribute_edges.tsv")
#'
#' # Standalone merge
#' merged_db <- merge_drugbank_hpo(drugbank, hpo)
#'
#' # Chained with other merges (any order)
#' full_db <- drugbank %>%
#' merge_drugbank_hpo(hpo) %>%
#' merge_drugbank_onsides(onsides) %>%
#' merge_drugbank_twosides(twosides)
#'
#' # --- Example: Find drugs targeting genes associated with seizures ---
#' seizure_drugs <- full_db$integrated_data$gene_phenotype_enriched %>%
#' filter(grepl("seizure", phenotype_name, ignore.case = TRUE)) %>%
#' filter(!is.na(drugbank_id)) %>%
#' select(drug_name, drugbank_id, gene_symbol, phenotype_name) %>%
#' distinct()
#' }
merge_drugbank_hpo <- function(db_object, hpo_db) {
# --- Step 0: Input Validation and Hub Detection ---
# Enables Pipe (%>%) and Chaining with other merge functions.
if ("drugbank" %in% names(db_object)) {
# CASE A: Input is an already-merged object (from another merge)
drugbank_db <- db_object$drugbank
merged_object <- db_object
} else {
# CASE B: Input is a raw DrugBank object
drugbank_db <- db_object
merged_object <- init_dvobject()
merged_object$drugbank <- db_object
attr(merged_object, "DrugBankDB") <- attr(drugbank_db, "original_db_info")
merged_object <- set_source_metadata(merged_object,
attr(drugbank_db, "source_metadata"))
}
# Validate the Hub — need cett$targets structure
if (!inherits(drugbank_db, "dvobject") ||
(!"cett" %in% names(drugbank_db))) {
stop("`db_object` must contain a valid DrugBank dvobject with `cett` data.",
call. = FALSE)
}
if (!"targets" %in% names(drugbank_db$cett)) {
stop("`db_object$cett` must contain `targets` data.",
call. = FALSE)
}
# Validate the Spoke
if (!inherits(hpo_db, "dvobject") ||
(!"associations" %in% names(hpo_db))) {
stop("`hpo_db` must be a valid dvobject from parseHPO() ",
"containing `associations`.", call. = FALSE)
}
# --- Step 1: Build the Gene-to-DrugBankID Bridge ---
#
# Path: polypeptides$general_information (has gene_name + target_id)
# → targets$general_information (has target_id + drugbank_id)
#
message("Building gene_name <-> DrugBank ID bridge via target_id...")
# Extract polypeptides (where gene_name lives)
polypeptides_gi <- .resolve_polypeptides_gi(drugbank_db)
if (is.null(polypeptides_gi) || (NROW(polypeptides_gi) == 0L)) {
stop("Could not locate polypeptides general_information table. ",
"Expected at: $cett$targets$polypeptides$general_information",
call. = FALSE)
}
# Extract targets general_information (where drugbank_id lives)
targets_gi <- .resolve_targets_gi(drugbank_db)
if (is.null(targets_gi) || (NROW(targets_gi) == 0L)) {
stop("Could not locate targets general_information table. ",
"Expected at: $cett$targets$general_information",
call. = FALSE)
}
# Build the bridge: gene_name → target_id → drugbank_id
gene_bridge <- polypeptides_gi %>%
dplyr::filter(!is.na(.data$gene_name) &
(trimws(.data$gene_name) != "")) %>%
dplyr::select("gene_name", "target_id",
polypeptide_name = "name",
"polypeptide_id") %>%
dplyr::distinct() %>%
dplyr::inner_join(
targets_gi %>%
dplyr::select("target_id", "drugbank_id") %>%
dplyr::distinct(),
by = "target_id"
) %>%
dplyr::mutate(gene_name_upper = toupper(trimws(.data$gene_name)))
message(sprintf(
" Bridge: %s gene-target-drug links\n %d unique genes | %d unique targets | %d unique drugs",
format(NROW(gene_bridge), big.mark = ","),
dplyr::n_distinct(gene_bridge$gene_name),
dplyr::n_distinct(gene_bridge$target_id),
dplyr::n_distinct(gene_bridge$drugbank_id)
))
# --- Step 2: Add Drug Names to Bridge ---
drug_name_lookup <- .resolve_drug_names(drugbank_db)
if (!is.null(drug_name_lookup) && (NROW(drug_name_lookup) > 0L)) {
gene_bridge <- gene_bridge %>%
dplyr::left_join(drug_name_lookup, by = "drugbank_id")
n_named <- sum(!is.na(gene_bridge$drug_name))
message(sprintf(" Drug names resolved for %d / %d bridge entries.",
n_named, NROW(gene_bridge)))
}
# --- Step 3: Enrich HPO Gene-Phenotype Associations ---
message("Enriching HPO gene-phenotype associations with DrugBank IDs...")
associations <- hpo_db$associations
hpo_enriched <- associations %>%
dplyr::mutate(gene_symbol_upper = toupper(trimws(.data$gene_symbol))) %>%
dplyr::left_join(
gene_bridge %>%
dplyr::select("gene_name_upper", "gene_name",
"target_id", "drugbank_id",
"drug_name", "polypeptide_name") %>%
dplyr::distinct(),
by = c("gene_symbol_upper" = "gene_name_upper"),
relationship = "many-to-many"
) %>%
dplyr::select(-dplyr::all_of("gene_symbol_upper"))
# Report match statistics
n_hpo_genes <- dplyr::n_distinct(associations$gene_symbol)
n_matched_genes <- hpo_enriched %>%
dplyr::filter(!is.na(.data$drugbank_id)) %>%
dplyr::pull("gene_symbol") %>%
unique() %>%
length()
n_enriched_rows <- hpo_enriched %>%
dplyr::filter(!is.na(.data$drugbank_id)) %>%
NROW()
message(sprintf(
paste0(" %d / %d HPO genes matched to DrugBank targets (%.1f%%)\n",
" Total enriched rows: %s (%s with drug links)"),
n_matched_genes, n_hpo_genes,
100 * n_matched_genes / max(n_hpo_genes, 1L),
format(NROW(hpo_enriched), big.mark = ","),
format(n_enriched_rows, big.mark = ",")
))
# --- Step 4: Assemble Final Merged Object ---
message("Assembling final merged object...")
# Store raw HPO spoke
merged_object$hpo <- hpo_db
# Initialize integrated_data if needed
if (is.null(merged_object$integrated_data)) {
merged_object$integrated_data <- list()
}
merged_object$integrated_data$gene_phenotype_enriched <- hpo_enriched
merged_object$integrated_data$DrugBank_GeneSymbol_Mapping <- gene_bridge
# --- Step 5: Metadata ---
attr(merged_object, "HPODB") <- attr(hpo_db, "original_db_info")
merged_object <- set_source_metadata(merged_object,
attr(hpo_db, "source_metadata"))
class(merged_object) <- unique(c("DrugBankHPODb", class(merged_object)))
message("Merge complete.")
merged_object
}
# =============================================================================
# Internal helpers for navigating the DrugBank dvobject structure
# =============================================================================
#' Resolve polypeptides general_information table
#'
#' Navigates: $cett$targets$polypeptides$general_information
#' Contains: gene_name, target_id, polypeptide_id, name, ...
#'
#' @param drugbank_db A DrugBank dvobject.
#' @return A tibble or NULL.
#' @keywords internal
.resolve_polypeptides_gi <- function(drugbank_db) {
# Primary path: $cett$targets$polypeptides$general_information
pp <- tryCatch(
drugbank_db$cett$targets$polypeptides$general_information,
error = function(e) NULL
)
if (is.data.frame(pp) && (NROW(pp) > 0L) &&
("gene_name" %in% names(pp)) &&
("target_id" %in% names(pp))) {
return(tibble::as_tibble(pp))
}
# Fallback: search for any table with both gene_name and target_id
.find_table_with_columns(drugbank_db,
required_cols = c("gene_name", "target_id"))
}
#' Resolve targets general_information table
#'
#' Navigates: $cett$targets$general_information
#' Contains: target_id, drugbank_id, name, organism, ...
#'
#' @param drugbank_db A DrugBank dvobject.
#' @return A tibble or NULL.
#' @keywords internal
.resolve_targets_gi <- function(drugbank_db) {
# Primary path: $cett$targets$general_information
tgi <- tryCatch(
drugbank_db$cett$targets$general_information,
error = function(e) NULL
)
if (is.data.frame(tgi) && (NROW(tgi) > 0L) &&
("target_id" %in% names(tgi)) &&
("drugbank_id" %in% names(tgi))) {
return(tibble::as_tibble(tgi))
}
# Fallback
.find_table_with_columns(drugbank_db,
required_cols = c("target_id", "drugbank_id"))
}
#' Resolve drug name lookup from a DrugBank dvobject
#'
#' Returns a two-column tibble: drugbank_id, drug_name.
#' Navigates: $drugs$general_information
#'
#' @param drugbank_db A DrugBank dvobject.
#' @return A tibble or NULL.
#' @keywords internal
.resolve_drug_names <- function(drugbank_db) {
# Primary path: $drugs$general_information
gi <- tryCatch(
drugbank_db$drugs$general_information,
error = function(e) NULL
)
if (is.data.frame(gi) && (NROW(gi) > 0L) &&
("drugbank_id" %in% names(gi)) &&
("name" %in% names(gi))) {
return(
gi %>%
dplyr::select("drugbank_id", drug_name = "name") %>%
dplyr::distinct()
)
}
# Fallback: flat $drugs as data frame
if (is.data.frame(drugbank_db$drugs) &&
("drugbank_id" %in% names(drugbank_db$drugs)) &&
("name" %in% names(drugbank_db$drugs))) {
return(
drugbank_db$drugs %>%
dplyr::select("drugbank_id", drug_name = "name") %>%
dplyr::distinct()
)
}
NULL
}
#' Generic fallback: search an object tree for a data frame with required columns
#'
#' @param obj A list (possibly nested).
#' @param required_cols Character vector of column names that must all be present.
#' @param max_depth Integer. Maximum recursion depth.
#' @return A tibble or NULL.
#' @keywords internal
.find_table_with_columns <- function(obj, required_cols, max_depth = 5L) {
if (max_depth <= 0L) return(NULL)
if (is.data.frame(obj) && (NROW(obj) > 0L)) {
if (all(required_cols %in% names(obj))) {
return(tibble::as_tibble(obj))
}
}
if (is.list(obj) && (!is.data.frame(obj))) {
for (nm in names(obj)) {
result <- .find_table_with_columns(obj[[nm]], required_cols,
max_depth = max_depth - 1L)
if (!is.null(result)) return(result)
}
}
NULL
}
Any scripts or data that you put into this service are public.
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.