data-raw/drug_utilities.R
In sigven/oncoPharmaDB: Molecularly targeted cancer drugs and biomarkers
## Read URL
#'
#' Function that retrieves PubChem compound properties using PubChem public user gateway (PUG)
#'
#' @param q_url Query URL
#' @return a data frame with the following columns of chemical compound properties:
#'
#'
readUrl <- function(q_url) {
out <- tryCatch(
{
readLines(con=q_url, warn=FALSE)
},
error=function(cond) {
# Choose a return value in case of error
return(NA)
},
warning=function(cond) {
# Choose a return value in case of warning
return(NULL)
},
finally={
#message(paste("Processed URL:", q_url))
#message("Some other message at the end")
}
)
return(out)
}
### NCI DRUG DISPLAY LABELS
process_nci_labels <- function(path_data_raw, overwrite = F) {
nci_thesaurus_labels_filepath = file.path(path_data_raw, "nci_thesaurus", "ThesaurusLabels.tsv")
nci_thesaurus_owl_filepath <- file.path(path_data_raw, "nci_thesaurus", "ThesaurusInferred.owl")
if(!is.null(nci_thesaurus_owl_filepath) &
file.exists(nci_thesaurus_owl_filepath) &
(!file.exists(nci_thesaurus_labels_filepath) | overwrite == T)){
system(paste0('/Users/sigven/miniconda3/bin/python data-raw/parse_owl_labels.py ',nci_thesaurus_owl_filepath,' > ',file.path(path_data_raw,'nci_thesaurus','ThesaurusLabels.tsv')))
}
nci_labels <- read.table(file = nci_thesaurus_labels_filepath, sep="\t",comment.char="",
stringsAsFactors = F,quote="",header=F) |>
magrittr::set_colnames(c('nci_t','nci_cd_name','cui')) |>
dplyr::filter(!is.na(nci_t))
return(nci_labels)
}
### CHEMBL-PUBCHEM COMPOUND CROSS-REFERENCE
get_chembl_pubchem_xref <- function(datestamp = '20241024',
chembl_release = "v34",
path_data_raw = NULL,
update = F){
chembl_pubchem_xref_fname <- file.path(
path_data_raw, "chembl",
paste0("chembl_pubchem_mapping_", datestamp,".txt.gz"))
if(!file.exists(chembl_pubchem_xref_fname) | update == T){
download.file(
"ftp://ftp.ebi.ac.uk/pub/databases/chembl/UniChem/data/wholeSourceMapping/src_id1/src1src22.txt.gz",
destfile = chembl_pubchem_xref_fname)
}
chembl_pubchem_compound_xref <- read.table(
gzfile(chembl_pubchem_xref_fname), header = F, skip = 1,
stringsAsFactors = F,sep="\t",quote="") |>
magrittr::set_colnames(c('molecule_chembl_id','pubchem_cid')) |>
dplyr::mutate(chembl_db_version = chembl_release)
return(chembl_pubchem_compound_xref)
}
### TARGETED ANTICANCER COMPOUNDS FROM OPEN TARGETS
get_otp_cancer_drugs <-
function(path_data_raw = NULL,
ot_version = "2024.09"){
cancer_terms <- list()
cancer_terms[['all']] <- phenOncoX::get_terms(
cache_dir = path_data_raw
)
cancer_terms[['efo']] <- cancer_terms[['all']]$records |>
dplyr::select(cui, cui_name, efo_id, primary_site) |>
dplyr::filter(!is.na(efo_id)) |>
dplyr::distinct() |>
dplyr::rename(disease_efo_id = efo_id)
fname <- paste0(path_data_raw,
paste0("/opentargets/opentargets_drugs_",
ot_version,".rds"))
ot_compounds <- as.data.frame(
readRDS(file = fname) |>
dplyr::select(target_genename,
target_symbol,
target_type,
target_ensembl_gene_id,
target_entrezgene,
disease_efo_id,
disease_efo_label,
drug_name,
drug_type,
drug_max_ct_phase,
drug_max_phase_indication,
molecule_chembl_id,
drug_action_type,
drug_moa,
drug_tradenames,
drug_synonyms,
drug_clinical_source,
drug_clinical_id,
drug_description,
drug_withdrawn,
drug_approved_indication,
drug_blackbox_warning,
drug_year_first_approval)) |>
dplyr::mutate(
disease_efo_id = stringr::str_replace_all(
disease_efo_id, "_", ":")) |>
dplyr::distinct() |>
dplyr::left_join(
cancer_terms[['efo']],
by = "disease_efo_id",
multiple = "all",
relationship = "many-to-many") |>
## general cancer ontology terms (neoplasm, cancer, carcinoma, squamous cell carcinoma)
dplyr::mutate(
drug_cancer_relevance =
dplyr::if_else(
!is.na(primary_site) |
disease_efo_id == 'EFO:0000616' |
disease_efo_id == 'EFO:0000311' |
disease_efo_id == 'EFO:0000313' |
disease_efo_id == 'MONDO:0004992' |
disease_efo_id == 'MONDO:0002120' |
disease_efo_id == 'EFO:0003826' |
disease_efo_id == 'EFO:0000707' |
disease_efo_id == 'EFO:0000228',
"by_cancer_condition_otp", "by_other_condition_otp")
) |>
dplyr::mutate(
drug_cancer_relevance = dplyr::if_else(
is.na(disease_efo_id) &
(drug_type == "Antibody" |
drug_type == "Unknown" |
drug_type == "Small molecule") &
!is.na(target_genename) &
stringr::str_detect(
tolower(target_genename),
paste0(
"(kinase|epidermal growth factor|tubulin|",
"oncogene|fibroblast growth factor|",
"dna topoisomerase|tnf receptor superfamily|",
"protein tyrosine phosphatase|",
"receptor tyrosine kinase|",
"platelet derived growth factor)"
)) &
(drug_action_type == "INHIBITOR" |
drug_action_type == "BINDING AGENT" |
drug_action_type == "ANTAGONIST"),
"by_cancer_target_otp",
as.character(drug_cancer_relevance)
)
) |>
dplyr::mutate(
drug_type = dplyr::if_else(
drug_type == "unknown",
"Unknown",
as.character(drug_type)
)
) |>
# dplyr::filter(
# !is.na(drug_action_type)
# ) |>
# dplyr::filter(
# !stringr::str_detect(
# tolower(drug_action_type),
# "allosteric|modulator|substrate|releasing|opener"
# )
# ) |>
dplyr::filter(
drug_type != "Cell" &
drug_type != "Enzyme" &
drug_type != "Gene" &
drug_type != "Oligonucleotide" &
drug_type != "Oligosaccharide"
)
cancer_indication_stats <- ot_compounds |>
dplyr::filter(
!is.na(drug_cancer_relevance) &
drug_cancer_relevance == "by_cancer_condition_otp") |>
dplyr::select(drug_name, disease_efo_label) |>
dplyr::filter(!is.na(disease_efo_label)) |>
dplyr::distinct() |>
dplyr::group_by(drug_name) |>
dplyr::summarise(n_cancer_indications = dplyr::n())
non_cancer_indication_stats <- ot_compounds |>
dplyr::filter(
!is.na(drug_cancer_relevance) &
drug_cancer_relevance == "by_other_condition_otp") |>
#dplyr::filter(drug_cancer_relevance == F) |>
dplyr::filter(!is.na(disease_efo_label)) |>
dplyr::select(drug_name, disease_efo_label,
drug_approved_indication) |>
dplyr::distinct() |>
dplyr::group_by(drug_name) |>
dplyr::summarise(n_noncancer_indications = dplyr::n(),
approved_non_cancer = paste(
unique(drug_approved_indication),
collapse = ";"
), .groups = "drop") |>
dplyr::mutate(approved_non_cancer = dplyr::if_else(
stringr::str_detect(approved_non_cancer,"TRUE"),
as.logical(TRUE),
as.logical(FALSE)
))
indication_stats <- cancer_indication_stats |>
dplyr::full_join(non_cancer_indication_stats, by = "drug_name") |>
dplyr::mutate(n_cancer_indications = dplyr::if_else(
is.na(n_cancer_indications),
as.numeric(0),
as.numeric(n_cancer_indications)
)) |>
dplyr::mutate(n_noncancer_indications = dplyr::if_else(
is.na(n_noncancer_indications),
as.numeric(0),
as.numeric(n_noncancer_indications)
)) |>
dplyr::mutate(drug_n_indications = n_cancer_indications +
n_noncancer_indications) |>
dplyr::mutate(drug_frac_cancer_indications = dplyr::if_else(
n_cancer_indications == 0,
as.numeric(0),
round(as.numeric(n_cancer_indications /
(n_cancer_indications + n_noncancer_indications)),
digits = 3)
)) |>
dplyr::mutate(drug_approved_noncancer = dplyr::if_else(
is.na(approved_non_cancer),
FALSE,
as.logical(approved_non_cancer)
)) |>
dplyr::select(drug_name,
drug_n_indications,
drug_approved_noncancer,
drug_frac_cancer_indications)
ot_compounds <- ot_compounds |>
dplyr::left_join(indication_stats, by = "drug_name") |>
dplyr::mutate(
drug_n_indications = dplyr::if_else(
is.na(drug_n_indications) |
drug_n_indications == "NA",
0,
as.integer(drug_n_indications)
)
)
targeted_compounds <- as.data.frame(
ot_compounds |>
dplyr::distinct() |>
dplyr::group_by_at(
dplyr::vars(-c(drug_clinical_id))) |>
dplyr::summarise(
drug_clinical_id = paste(
unique(drug_clinical_id),
collapse = ","),
drug_clinical_source = paste(
unique(sort(drug_clinical_source)),
collapse = ","),
.groups = "drop") |>
dplyr::ungroup() |>
dplyr::mutate(
drug_name_lc = tolower(drug_name))
)
## adjust max ct phase
drugs_with_max_phase_adj <- as.data.frame(
targeted_compounds |>
#dplyr::filter(drug_cancer_relevance == T) |>
dplyr::filter(!is.na(drug_max_ct_phase) &
!is.na(drug_max_phase_indication)) |>
dplyr::group_by(drug_name, molecule_chembl_id) |>
dplyr::summarise(drug_max_ct_phase = max(drug_max_phase_indication),
.groups = "drop")
)
targeted_compounds <- as.data.frame(targeted_compounds |>
dplyr::select(-drug_max_ct_phase) |>
dplyr::left_join(
drugs_with_max_phase_adj,
by = c("drug_name","molecule_chembl_id"),
multiple = "all", relationship = "many-to-many")
)
targeted_cancer_compounds <- targeted_compounds |>
dplyr::filter(
!is.na(drug_cancer_relevance) &
(drug_cancer_relevance == "by_cancer_condition_otp" |
drug_cancer_relevance == "by_cancer_target_otp")) |>
dplyr::mutate(drug_n_indications = dplyr::if_else(
is.na(drug_n_indications),as.numeric(0),
as.numeric(drug_n_indications)
)) |>
dplyr::mutate(drug_frac_cancer_indications = dplyr::if_else(
is.na(drug_frac_cancer_indications),as.numeric(0),
as.numeric(drug_frac_cancer_indications)
)) |>
dplyr::mutate(drug_approved_noncancer = dplyr::if_else(
is.na(drug_approved_noncancer),as.logical(FALSE),
as.logical(drug_approved_noncancer)
)) |>
dplyr::filter(drug_withdrawn == F)
targeted_noncancer_compounds <- targeted_compounds |>
# dplyr::filter(
# !is.na(drug_cancer_relevance) &
# drug_cancer_relevance == "by_other_condition_otp") |>
dplyr::anti_join(
dplyr::select(
targeted_cancer_compounds, molecule_chembl_id
), by = "molecule_chembl_id"
) |>
dplyr::filter(drug_withdrawn == F)
targeted_noncancer_compounds$drug_name_lc <- NULL
targeted_cancer_compounds$drug_name_lc <- NULL
ot_recs <- dplyr::bind_rows(
targeted_cancer_compounds,
targeted_noncancer_compounds) |>
dplyr::distinct()
return(ot_recs)
}
get_atc_drug_classification <- function(
path_data_raw = NULL){
atc_codes <- as.data.frame(
readr::read_csv(
file = file.path(
path_data_raw, "atc",
"ATC.csv.gz"),
show_col_types = F)) |>
janitor::clean_names() |>
dplyr::mutate(class_id = stringr::str_replace(
class_id, "http://purl.bioontology.org/ontology/ATC/",""
)) |>
dplyr::mutate(parents = stringr::str_replace(
parents, "http://purl.bioontology.org/ontology/ATC/",""
)) |>
dplyr::mutate(
is_drug_class = dplyr::if_else(
is.na(is_drug_class),
FALSE,
TRUE
)
) |>
dplyr::select(class_id, preferred_label, synonyms, obsolete,
parents, atc_level, is_drug_class) |>
dplyr::arrange(atc_level, class_id, parents)
#dplyr::filter(stringr::str_detect(class_id,"^L"))
atc_level1 <- atc_codes |>
dplyr::filter(atc_level == 2) |>
dplyr::rename(atc_level1 = preferred_label,
atc_code_level1 = class_id) |>
dplyr::select(atc_code_level1, atc_level1)
atc_level2 <- atc_codes |>
dplyr::filter(atc_level == 3) |>
dplyr::rename(atc_level2 = preferred_label,
atc_code_level2 = class_id, parent_code = parents) |>
dplyr::select(atc_code_level2, atc_level2, parent_code)
atc_level3 <- atc_codes |>
dplyr::filter(atc_level == 4) |>
dplyr::rename(atc_level3 = preferred_label,
atc_code_level3 = class_id,
parent_code = parents) |>
dplyr::select(atc_code_level3,
atc_level3, parent_code)
atc_level4 <- atc_codes |>
dplyr::filter(atc_level == 5) |>
dplyr::rename(atc_level4 = preferred_label,
atc_code_level4 = class_id,
parent_code = parents) |>
dplyr::select(atc_code_level4,
atc_level4,
parent_code)
atc_drug_classification <- atc_level1 |>
dplyr::left_join(
atc_level2, by = c("atc_code_level1" = "parent_code")) |>
dplyr::left_join(
atc_level3, by = c("atc_code_level2" = "parent_code")
) |>
dplyr::left_join(
atc_level4, by = c("atc_code_level3" = "parent_code")
) |>
dplyr::rename(atc_drug_entry = atc_level4) |>
dplyr::mutate(atc_level2 = stringr::str_replace(
atc_level2,
"in ATC|, ANTINEOPLASTIC DRUGS|, ANTINEOPLASTIC AND IMMUNOMODULATING AGENTS",""
)) |>
dplyr::mutate(atc_level3 = stringr::str_replace(
atc_level3,
"in ATC",""
)) |>
dplyr::mutate(atc_level2 = stringr::str_replace(
atc_level2, "AND ANTIBODY DRUG CONJUGATES","AND ADCs"
)) |>
dplyr::mutate(atc_level3 = dplyr::case_when(
atc_code_level3 == "L03AX" ~ "Other immunostimulants",
atc_code_level3 == "L01CD" ~ "Taxanes",
atc_code_level3 == "L04AX" ~ "Other immunosuppressants",
atc_code_level3 == "L04AB" ~ "TNF-alpha inhibitors",
atc_code_level3 == "L02BX" ~ "Other hormone antagonists",
atc_code_level3 == "L02AE" ~ "GnRH analogs",
atc_code_level3 == "L01XK" ~ "PARP inhibitors",
atc_code_level3 == "L01XX" ~ "Other antineoplastic agents",
atc_code_level3 == "L02AX" ~ "Other hormones for endocrine therapy",
atc_code_level3 == "L02BB" ~ "Anti-androgens",
atc_code_level3 == "L01XH" ~ "HDAC inhibitors",
atc_code_level3 == "L01XG" ~ "Proteasome inhibitors",
atc_code_level3 == "L01XA" ~ "Platinum compounds",
atc_code_level3 == "L01FX" ~ "Other MaBs and ADCs",
atc_code_level3 == "L01FF" ~ "PD-1/PDL-1 inhibitors",
atc_code_level3 == "L01FG" ~ "VEGFR inhibitors",
atc_code_level3 == "L01FE" ~ "EGFR inhibitors",
atc_code_level3 == "L01FD" ~ "HER2 inhibitors",
atc_code_level3 == "L01FC" ~ "CD38 inhibitors",
atc_code_level3 == "L01FB" ~ "CD22 inhibitors",
atc_code_level3 == "L01FA" ~ "CD20 inhibitors",
atc_code_level3 == "L01EX" ~ "Other protein kinase inhibitors",
atc_code_level3 == "L01EN" ~ "FGFR inhibitors",
atc_code_level3 == "L01EM" ~ "PI3K inhibitors",
atc_code_level3 == "L01EL" ~ "BTK inhibitors",
atc_code_level3 == "L01EK" ~ "VEGFR inhibitors",
atc_code_level3 == "L01EJ" ~ "JAK inhibitors",
atc_code_level3 == "L01EH" ~ "HER2 inhibitors",
atc_code_level3 == "L01EG" ~ "mTOR inhibitors",
atc_code_level3 == "L01EF" ~ "CDK inhibitors",
atc_code_level3 == "L01EE" ~ "MEK inhibitors",
atc_code_level3 == "L01ED" ~ "ALK inhibitors",
atc_code_level3 == "L01EC" ~ "BRAF inhibitors",
atc_code_level3 == "L01EB" ~ "EGFR inhibitors",
atc_code_level3 == "L01EA" ~ "BCR-ABL inhibitors",
atc_code_level3 == "L01DC" ~ "Other cytotoxic antibiotics",
atc_code_level3 == "L01CX" ~ "Other plant alkaloids and natural products",
atc_code_level3 == "L01CE" ~ "TOP1 inhibitors",
atc_code_level3 == "L01CC" ~ "Colchicine derivatives",
atc_code_level3 == "L01CB" ~ "Podophyllotoxin derivatives",
atc_code_level3 == "L01CA" ~ "Vinca alkaloids and analogues",
atc_code_level3 == "L01BC" ~ "Pyrimidine analogues",
atc_code_level3 == "L01BB" ~ "Purine analogues",
atc_code_level3 == "L01BA" ~ "Folic acid analogues",
atc_code_level3 == "L01AX" ~ "Other alkylating agents",
atc_code_level3 == "L01XB" ~ "Methylhydrazines",
atc_code_level3 == "L01AG" ~ "Epoxides",
atc_code_level3 == "L01AD" ~ "Nitrosoureas",
atc_code_level3 == "L01AC" ~ "Ethylene imines",
TRUE ~ as.character(atc_level3)
)) |>
dplyr::select(-atc_code_level4) |>
dplyr::distinct()
#dplyr::mutate(atc_drug_entry = dplyr::if_else(
#atc_code_level3 == "L01EX" |
#atc_code_level3 == "L01FX" |
#atc_code_level3 == "L01XX",
#as.character(NA),
#as.character(atc_drug_entry)
#))
atc_custom <- as.data.frame(
readr::read_csv(
file.path(
path_data_raw,
"atc",
"custom_classification.csv"
), col_names = T, show_col_types = F, na = c(".")
)
)
atc_drug_classification <- atc_drug_classification |>
dplyr::bind_rows(atc_custom) |>
dplyr::arrange(atc_code_level3) |>
dplyr::distinct() |>
dplyr::mutate(atc_code_level3 = dplyr::case_when(
## JAK inhibitors (assigned to both L01EJ and L04AF,
## stick to L01EJ for consistency with other kinase inhibitors)
atc_code_level3 == "L04AF" ~ "L01EJ",
## mTOR inhibitors (assigned to both L04AH and L01EG,
## stick to L01EG for consistency with other kinase inhibitors)
atc_code_level3 == "L04AH" ~ "L01EG",
atc_drug_entry == "ipilimumab" |
atc_drug_entry == "tremelimumab" ~ "L01FXA",
atc_drug_entry == "epcoritamab" |
atc_drug_entry == "glofitamab" |
atc_drug_entry == "mosunetuzumab" |
atc_drug_entry == "talquetamab" |
atc_drug_entry == "teclistamab" ~ "L01FXE",
TRUE ~ as.character(atc_code_level3)
)) |>
dplyr::mutate(atc_level3 = dplyr::case_when(
atc_code_level3 == "L01EJ" ~ "JAK inhibitors",
atc_code_level3 == "L01EG" ~ "mTOR inhibitors",
atc_drug_entry == "ipilimumab" |
atc_drug_entry == "tremelimumab" ~ "Other ICIs - CTLA4 inhibitors",
atc_drug_entry == "epcoritamab" |
atc_drug_entry == "glofitamab" |
atc_drug_entry == "mosunetuzumab" |
atc_drug_entry == "talquetamab" |
atc_drug_entry == "teclistamab" ~ "T-cell Engagers",
TRUE ~ as.character(atc_level3)
))
return(atc_drug_classification)
}
## NCI THESAURUS CANCER DRUGS/TREATMENTS
get_nci_drugs <- function(nci_db_release = nci_db_release,
overwrite = F,
path_data_raw = NULL,
path_data_processed = NULL){
nci_antineo_thesaurus <- NULL
nci_drugs <- NULL
nci_ftp_base <- paste0(
"https://evs.nci.nih.gov/ftp1/NCI_Thesaurus/archive/",
nci_db_release,
"_Release/")
nci_thesaurus_files <- list()
nci_thesaurus_files[['flat']] <-
paste0("Thesaurus_", nci_db_release,".FLAT.zip")
nci_thesaurus_files[['owl']] <-
paste0("Thesaurus_", nci_db_release,".OWL.zip")
nci_thesaurus_files[['inf_owl']] <-
paste0("ThesaurusInf_", nci_db_release,".OWL.zip")
options(timeout = 50000)
for (elem in c('flat','owl','inf_owl')) {
remote_file <- paste0(
nci_ftp_base, nci_thesaurus_files[[elem]])
local_file <- file.path(
path_data_raw,"nci_thesaurus",
nci_thesaurus_files[[elem]])
if (!file.exists(local_file)) {
download.file(
url = remote_file,
destfile = local_file,
quiet = T)
system(paste0('unzip -d ',
file.path(path_data_raw, "nci_thesaurus"),
' -o -u ',local_file))
}
}
antineo_agents_url <-
'https://evs.nci.nih.gov/ftp1/NCI_Thesaurus/Drug_or_Substance/Antineoplastic_Agent.txt'
antineo_agents_local <-
file.path(
path_data_raw,
"nci_thesaurus",
"Antineoplastic_Agent.txt")
if (!file.exists(antineo_agents_local)) {
download.file(
url = antineo_agents_url,
destfile = antineo_agents_local, quiet = T)
}
if(!file.exists(
file.path(
path_data_processed,
"nci_thesaurus",
"nci_treatment_thesaurus_antineo_v2.rds")) | overwrite == T){
sorafenib_definition <-
'A synthetic compound targeting growth signaling and angiogenesis. Sorafenib blocks the enzyme RAF kinase, a critical component of the RAF/MEK/ERK signaling pathway that controls cell division and proliferation; in addition, sorafenib inhibits the VEGFR-2/PDGFR-beta signaling cascade, thereby blocking tumor angiogenesis.'
nci_display_labels <- process_nci_labels(
path_data_raw = path_data_raw,
overwrite = overwrite)
## Agents/compounds marked as antineplastic according to NCI
nci_antineo_agents <-
read.table(file = file.path(path_data_raw,"nci_thesaurus","Antineoplastic_Agent.txt"),
header = T, stringsAsFactors = F, comment.char = "",
quote = "", sep = "\t") |>
janitor::clean_names() |>
dplyr::rename(nci_t = code) |>
dplyr::select(nci_t) |>
dplyr::mutate(antineoplastic_agent = TRUE) |>
dplyr::distinct()
## parse all entries in nci thesaurus where the
## semantic concept type is treatment-related
nci_antineo_thesaurus_raw <- as.data.frame(
read.table(file = file.path(path_data_raw, "nci_thesaurus", "Thesaurus.txt"),
header = F, stringsAsFactors = F, sep="\t",
comment.char="", quote = "") |>
dplyr::rename(nci_t = V1, nci_concept_name = V2,
nci_t_parent = V3, concept_synonym = V4,
nci_concept_definition = V5,
nci_cd_name = V6, nci_concept_status = V7,
nci_concept_semantic_type = V8) |>
dplyr::filter(
stringr::str_detect(
nci_concept_semantic_type,
"Chemical|Substance|Therapeutic|Drug|Immunologic")) |>
dplyr::filter(
nci_concept_semantic_type != "Drug Delivery Device" &
nci_concept_semantic_type != "Chemical Viewed Functionally"
) |>
dplyr::filter(
!stringr::str_detect(
tolower(nci_concept_definition), "surgical"
)
) |>
dplyr::filter(
nci_concept_semantic_type != "Therapeutic or Preventive Procedure" |
(nci_concept_semantic_type == "Therapeutic or Preventive Procedure" &
stringr::str_detect(
tolower(concept_synonym), "/"
) &
!stringr::str_detect(
tolower(nci_concept_definition), "regimen"
)
)
) |>
dplyr::left_join(
nci_antineo_agents, by = c("nci_t"),
multiple = "all", relationship = "many-to-many") |>
dplyr::select(-c(nci_concept_name, nci_cd_name)) |>
dplyr::mutate(nci_concept_synonym_all = concept_synonym) |>
tidyr::separate_rows(concept_synonym,sep="\\|") |>
dplyr::mutate(nci_concept_synonym = tolower(concept_synonym)) |>
dplyr::select(-concept_synonym) |>
dplyr::filter(
nci_t != 'C147908' & nci_t != 'C71622') |> ## Hormone Therapy Agent
dplyr::mutate(
nci_concept_semantic_type =
dplyr::if_else(
nci_t == 'C61948','Pharmacologic Substance',
nci_concept_semantic_type)) |> ## redundant Sorafenib entries
dplyr::mutate(
nci_concept_definition =
dplyr::if_else(
nci_t == 'C61948', sorafenib_definition,
nci_concept_definition)) |> ## redundant Sorafenib entries
dplyr::distinct() |>
dplyr::left_join(
nci_display_labels, by = c("nci_t"),
multiple = "all", relationship = "many-to-many") |>
dplyr::filter(
!(nci_t == "C1806" & nci_concept_synonym == "gemtuzumab")) |>
dplyr::filter(
!(nci_t == "C405" & nci_concept_synonym == "ctx")) |>
dplyr::mutate(nci_db_version = nci_db_release) |>
dplyr::filter(!is.na(nci_cd_name)) |>
dplyr::filter(!stringr::str_detect(
tolower(nci_concept_definition), "coronavirus")) |>
dplyr::filter(!stringr::str_detect(
nci_concept_synonym_all,
"SARS-CoV-2|COVID-19|CoV-19|Coronary|Corona|Covid-19|covid-19")) |>
#))
dplyr::mutate(antineoplastic_agent = dplyr::if_else(
is.na(antineoplastic_agent),
as.logical(TRUE),
as.logical(antineoplastic_agent)
)) |>
dplyr::filter(!stringr::str_detect(nci_cd_name," (Gel|Oil|Cream|Seed|Block|Field|Supplement|Factor)$")) |>
dplyr::filter(!stringr::str_detect(nci_cd_name,"(Vaccination|Lotion|Therapeutic Heat|Procedure|Rehabilitation|Prevention|Rinse)$")) |>
dplyr::filter(!stringr::str_detect(nci_cd_name,"(Epitope|Exract|Influenza|Ginseng|Ointment|Management|Injection|Tool)$")) |>
dplyr::filter(!stringr::str_detect(nci_cd_name,"^(Vitamin A Compound|Inactivated Poliovirus|Antineoplastic Immune Cell|Topical)")) |>
dplyr::filter(!stringr::str_detect(nci_cd_name,"^(Sheng-Yu|Ginseng|Dry Cleaning|Boost|Tobacco|Microwave)")) |>
dplyr::filter(!stringr::str_detect(nci_cd_name,"^(Blood|Interruption of|Gum Arabic|Vaginal Cylinder|Laser Ablation|Wheatgrass)")) |>
dplyr::filter(!stringr::str_detect(nci_cd_name,"^(Chemical Challenge|Prevention of|Magic Mouthwash|Wood Dust|Soot|Cocaine)")) |>
dplyr::filter(!stringr::str_detect(nci_cd_name,"^(Antibody|Antigen|Antioxidant|Vaccination|Acetate|Antiserum|Asbestos|Aspirate|Autoantigen|Cytokine)$")) |>
dplyr::filter(!stringr::str_detect(nci_cd_name," Spray| Extract| Antidiabetic| Implant|(Green Tea|Living Healthy|Pollutant|Probe|Protective Agent|Supportive Care|Caffe)"))
#dplyr::filter(
# !stringr::str_detect(
# tolower(nci_concept_definition),
# "chinese |antidiabet|diabetes|antidepress|analgesic|pulmonary edema|nutritional|human carcinogen|anesthetic|nonsedating|sedative|antihyper|antiinflamma|antiarrythm|antiangin|antihist|muscle|neurotransmitter"))
)
nci_antineo_thesaurus <- nci_antineo_thesaurus_raw
## add apatinib as an alias (is missing in NCI thesaurus)
apatinib_alias_entry <- nci_antineo_thesaurus |>
dplyr::filter(nci_concept_synonym == "rivoceranib") |>
dplyr::mutate(nci_concept_synonym = "apatinib") |>
dplyr::mutate(nci_concept_synonym_all =
paste(nci_concept_synonym_all,"Apatinib",sep="|"))
## add nab-paclitaxel as an alias (is missing in NCI thesaurus)
paclitaxel_entry <- nci_antineo_thesaurus |>
dplyr::filter(nci_cd_name == "Paclitaxel") |>
dplyr::select(nci_t, nci_t_parent, nci_concept_definition,
nci_concept_status, nci_concept_semantic_type,
nci_concept_synonym_all, cui) |>
dplyr::distinct()
nab_paclitaxel_entry <- nci_antineo_thesaurus |>
dplyr::filter(nci_cd_name == "Nab-paclitaxel")
nab_paclitaxel_entry$nci_t <- paclitaxel_entry$nci_t
nab_paclitaxel_entry$nci_t_parent <- paclitaxel_entry$nci_t_parent
nab_paclitaxel_entry$cui <- paclitaxel_entry$cui
nab_paclitaxel_entry$nci_concept_definition <- paclitaxel_entry$nci_concept_definition
nab_paclitaxel_entry$nci_concept_status <- paclitaxel_entry$nci_concept_status
nab_paclitaxel_entry$nci_concept_semantic_type <- paclitaxel_entry$nci_concept_semantic_type
nab_paclitaxel_entry$nci_cd_name <- "Paclitaxel"
aliases_paclitaxel <- stringr::str_split(
paclitaxel_entry$nci_concept_synonym_all, "\\|")[[1]]
aliases_nab_paclitaxel <- stringr::str_split(
nab_paclitaxel_entry$nci_concept_synonym_all[1], "\\|")[[1]]
all_paclitaxel_aliases <-
paste(sort(unique(c(aliases_paclitaxel, aliases_nab_paclitaxel))),
collapse="|")
nab_paclitaxel_entry$nci_concept_synonym_all <-
all_paclitaxel_aliases
paclitaxel_entries <- nci_antineo_thesaurus |>
dplyr::filter(nci_cd_name == "Paclitaxel")
paclitaxel_entries$nci_concept_synonym_all <-
all_paclitaxel_aliases
nci_antineo_thesaurus <- nci_antineo_thesaurus |>
dplyr::filter(nci_cd_name != "Paclitaxel" &
nci_cd_name != "Nab-paclitaxel") |>
dplyr::bind_rows(nab_paclitaxel_entry) |>
dplyr::bind_rows(paclitaxel_entries) |>
dplyr::bind_rows(apatinib_alias_entry) |>
dplyr::arrange(nci_cd_name) |>
dplyr::rename(nci_drug_name = nci_concept_synonym) |>
dplyr::filter(!(nci_cd_name == "Sorafenib Tosylate" &
nci_drug_name == "sorafenib")) |>
dplyr::select(-c(nci_t_parent,cui)) |>
dplyr::distinct()
nci_drugs <- list()
## NCI anticancer drugs (non-targeted) - lacking compound identifier (CHEMBL)
nci_drugs[['no_chembl_id']] <- nci_antineo_thesaurus |>
dplyr::select(nci_t,
nci_concept_definition,
nci_cd_name,
nci_drug_name,
nci_concept_synonym_all) |>
dplyr::filter(!stringr::str_detect(
nci_drug_name,
"(material|medication|medicine|medicinal|medical|^was |^agent |no therapy|-line therapy)"
)) |>
dplyr::filter(!stringr::str_detect(
tolower(nci_cd_name),
"(material|medication|medicine|medicinal|medical|^was |^agent |no therapy|-line therapy)"
)) |>
## remove general drug terms ("XX Agent" etc)
dplyr::mutate(num_words = stringr::str_count(
nci_cd_name, " ") + 1) |>
dplyr::filter(
!(num_words <= 4 &
stringr::str_detect(tolower(nci_cd_name), " agent") &
!stringr::str_detect(nci_cd_name, "[0-9]{1,}")
)) |>
dplyr::select(-num_words) |>
dplyr::distinct()
saveRDS(nci_drugs, file = file.path(
path_data_processed, "nci_thesaurus",
"nci_treatment_thesaurus_antineo_v2.rds"))
}else{
nci_drugs <- readRDS(
file = file.path(
path_data_processed, "nci_thesaurus",
"nci_treatment_thesaurus_antineo_v2.rds"))
}
return(nci_drugs)
}
merge_nci_opentargets <- function(
drug_sets = NULL,
path_data_raw = NULL){
## X-ref Open Targets and NCI by drug name
otp_drugs_all <- drug_sets[['otp']] |>
dplyr::mutate(drug_name_lc = tolower(drug_name)) |>
dplyr::left_join(
dplyr::filter(
drug_sets[['nci']][['no_chembl_id']],
!is.na(nci_drug_name)),
by = c("drug_name_lc" = "nci_drug_name"),
multiple = "all", relationship = "many-to-many") |>
dplyr::select(-c("drug_name_lc")) |>
dplyr::distinct() |>
dplyr::mutate(opentargets = TRUE)
## list all drug aliases in targeted drugs from OTP
all_drug_aliases <- otp_drugs_all |>
tidyr::separate_rows(drug_synonyms, sep = "\\|") |>
dplyr::select(drug_synonyms) |>
dplyr::mutate(nci_drug_name = tolower(drug_synonyms)) |>
dplyr::distinct()
## Identify drugs in NCI that are in Open Targets
found_drugs <- drug_sets[['nci']][['no_chembl_id']] |>
dplyr::semi_join(all_drug_aliases,
by = "nci_drug_name") |>
dplyr::select(nci_t, nci_drug_name) |>
dplyr::distinct()
nci_missing <- drug_sets[['nci']][['no_chembl_id']] |>
dplyr::anti_join(found_drugs,
by = "nci_t") |>
dplyr::select(-nci_drug_name) |>
dplyr::mutate(opentargets = F) |>
dplyr::distinct()
## ignore some drugs
custom_name_ignore <- as.data.frame(readr::read_tsv(
file = "data-raw/drug_names_ignore.tsv",
col_names = F, show_col_types = F))
colnames(custom_name_ignore) <- c("nci_cd_name")
## ignore some molecule identifiers (mislabeled)
custom_chembl_ignore <- as.data.frame(readr::read_tsv(
file = "data-raw/molecule_chembl_ids_ignore.tsv",
col_names = F, show_col_types = F))
colnames(custom_chembl_ignore) <- c("molecule_chembl_id")
all_cancer_drugs <- otp_drugs_all |>
dplyr::bind_rows(nci_missing) |>
dplyr::anti_join(
custom_chembl_ignore,
by = "molecule_chembl_id") |>
dplyr::distinct() |>
dplyr::mutate(
nci_cd_name =
dplyr::if_else(
is.na(nci_cd_name) &
!stringr::str_detect(drug_name,"[0-9]"),
Hmisc::capitalize(tolower(drug_name)),
nci_cd_name)) |>
dplyr::mutate(
nci_cd_name = dplyr::if_else(
is.na(nci_cd_name) &
stringr::str_detect(drug_name,"[0-9]"),
drug_name,
nci_cd_name)) |>
dplyr::mutate(
nci_cd_name = dplyr::if_else(
!is.na(nci_cd_name) &
!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL535",
"Sunitinib",
nci_cd_name)) |>
dplyr::mutate(
nci_cd_name = dplyr::if_else(
!is.na(drug_name) &
!is.na(nci_cd_name) &
!is.na(molecule_chembl_id) &
(molecule_chembl_id == "CHEMBL3545427" |
molecule_chembl_id == "CHEMBL5315068" |
molecule_chembl_id == "CHEMBL2109325" |
molecule_chembl_id == "CHEMBL5316170" |
molecule_chembl_id == "CHEMBL31965" |
molecule_chembl_id == "CHEMBL5314999" |
molecule_chembl_id == "CHEMBL2109402" |
molecule_chembl_id == "CHEMBL1790041" |
molecule_chembl_id == "CHEMBL4650278" |
molecule_chembl_id == "CHEMBL3545401"),
stringr::str_to_title(drug_name),
nci_cd_name)) |>
dplyr::mutate(drug_name = dplyr::if_else(
is.na(drug_name),
nci_cd_name,
drug_name)) |>
dplyr::anti_join(
custom_name_ignore,
by = "nci_cd_name")
rownames(all_cancer_drugs) <- NULL
salt_patterns <-
readr::read_tsv(
file.path(path_data_raw, "salts.tsv"),
show_col_types = F, col_names = F)
salt_patterns_regex <- paste0(
"( (",
paste(salt_patterns$X1, collapse="|"),
"))$")
salt_forms <- all_cancer_drugs |>
dplyr::filter(
stringr::str_detect(
tolower(nci_cd_name),
salt_patterns_regex)) |>
dplyr::filter(!is.na(opentargets_version)) |>
dplyr::mutate(tradename = stringr::str_replace(
nci_cd_name,
salt_patterns_regex,
"")) |>
dplyr::select(tradename, nci_cd_name) |>
dplyr::distinct() |>
dplyr::mutate(is_salt = T) |>
dplyr::inner_join(
dplyr::select(all_cancer_drugs, nci_cd_name),
by = c("tradename" = "nci_cd_name"),
multiple = "all", relationship = "many-to-many") |>
dplyr::distinct() |>
dplyr::select(-tradename)
all_cancer_drugs <- all_cancer_drugs |>
dplyr::left_join(
salt_forms,
by = "nci_cd_name",
multiple = "all",
relationship = "many-to-many") |>
dplyr::mutate(is_salt = dplyr::if_else(
is.na(is_salt),
as.logical(FALSE),
as.logical(is_salt)
)) |>
dplyr::distinct()
## antibody drug conjugates
adc_candidates <- all_cancer_drugs |>
dplyr::filter(
(!is.na(nci_cd_name) &
stringr::str_detect(tolower(nci_cd_name), "mab ")) |
stringr::str_detect(
nci_concept_definition, "ADC|antibody(-| )drug conjugate")) |>
dplyr::mutate(is_adc = TRUE) |>
dplyr::mutate(is_adc = dplyr::if_else(
stringr::str_detect(nci_concept_definition, "ADCC") &
!(stringr::str_detect(
tolower(nci_concept_definition),
"(antibody(-| )drug conjugate)") &
(is.na(drug_type) |
drug_type == "Small molecule" |
drug_type == "Antibody")),
as.logical(FALSE),
as.logical(is_adc)
)) |>
dplyr::mutate(is_adc = dplyr::if_else(
is_adc == F &
!is.na(nci_cd_name) &
stringr::str_detect(
tolower(nci_cd_name), "mab ") &
stringr::str_detect(
tolower(nci_cd_name), "(tin|ine|tan|can|tox)$"
),
TRUE,
as.logical(is_adc)
)) |>
dplyr::filter(!is.na(nci_cd_name)) |>
dplyr::select(nci_cd_name, is_adc) |>
dplyr::distinct()
all_cancer_drugs <- all_cancer_drugs |>
dplyr::left_join(
adc_candidates,
by = "nci_cd_name",
multiple = "all",
relationship = "many-to-many") |>
dplyr::mutate(is_adc = dplyr::if_else(
is.na(is_adc),
as.logical(FALSE),
as.logical(is_adc)
)) |>
dplyr::filter(drug_withdrawn == F | is.na(drug_withdrawn)) |>
dplyr::mutate(nci_concept_definition = dplyr::if_else(
!is.na(nci_concept_definition) &
nchar(nci_concept_definition) == 0,
as.character(NA),
as.character(nci_concept_definition)
)) |>
dplyr::mutate(drug_n_indications = dplyr::if_else(
is.na(drug_n_indications),as.numeric(0),
as.numeric(drug_n_indications)
)) |>
dplyr::mutate(drug_frac_cancer_indications = dplyr::if_else(
is.na(drug_frac_cancer_indications),as.numeric(0),
as.numeric(drug_frac_cancer_indications)
)) |>
dplyr::mutate(drug_approved_noncancer = dplyr::if_else(
is.na(drug_approved_noncancer),as.logical(FALSE),
as.logical(drug_approved_noncancer)
)) |>
dplyr::mutate(drug_cancer_relevance = dplyr::if_else(
is.na(drug_cancer_relevance) &
opentargets == F,
"by_nci",
as.character(drug_cancer_relevance)
)) |>
dplyr::distinct()
return(all_cancer_drugs)
}
map_curated_targets <- function(gene_info = NULL,
path_data_raw = NULL,
drug_df = NULL){
drug_df$target_entrezgene <- as.numeric(
drug_df$target_entrezgene
)
drug_target_patterns <-
read.table(file = file.path(
path_data_raw,
"custom_drug_target_regex_nci.tsv"),
sep = "\t", header = T,
stringsAsFactors = F, quote = "") |>
dplyr::inner_join(
gene_info, by = "symbol", multiple = "all",
relationship = "many-to-many") |>
dplyr::distinct()
drugname_suffix <-
paste0("(ab|al|an|at|cl|co|da|de|dy|ea|ed|el|en|er|es|ex|fa|ib|ic|",
"id|il|im|in|ir|is|le|ls|lt|me|na|ne|ns|nt|od|ol|on|or|pt|",
"py|ra|rd|re|rm|rt|se|ta|te|ts|ue|um|us|yl)$")
drug_inhibitor_regex <-
paste0(drugname_suffix,"|^(anti-|inhibitor of)|inhibitor")
all_inhibitors_no_target <- drug_df |>
dplyr::filter(is.na(target_symbol)) |>
dplyr::filter(
stringr::str_detect(
tolower(nci_cd_name),
"inhibitor|antagonist|antibody|blocker|sepantronium| mimetic") |
stringr::str_detect(
tolower(nci_cd_name),
drug_inhibitor_regex) |
(stringr::str_detect(nci_concept_definition,"KRAS") &
stringr::str_detect(nci_concept_definition,"inhibitor"))) |>
dplyr::filter(!stringr::str_detect(
nci_cd_name,
" CAR T|(T|t)herapy|SARS-CoV-2|( (R|r)egimen|(A|a)cid|Hydrochloride)$")) |>
dplyr::filter(!stringr::str_detect(
nci_concept_definition,
"SARS-CoV-2")) |>
dplyr::filter(nchar(nci_concept_definition) > 0)
custom_nci_targeted_drugs <- data.frame()
for(i in 1:nrow(drug_target_patterns)){
pattern <- drug_target_patterns[i, "pattern"]
target_symbol <- drug_target_patterns[i, "symbol"]
target_genename <- drug_target_patterns[i, "genename"]
target_entrezgene <- drug_target_patterns[i, "target_entrezgene"]
target_type <- drug_target_patterns[i, "target_type"]
target_ensembl_gene_id <- drug_target_patterns[i, "target_ensembl_gene_id"]
#target_uniprot_id <- drug_target_patterns[i, "target_uniprot_id"]
hits <- all_inhibitors_no_target |>
dplyr::filter(stringr::str_detect(
nci_cd_name,
pattern = pattern) |
(stringr::str_detect(
tolower(nci_cd_name),
drug_inhibitor_regex) &
stringr::str_detect(
nci_concept_definition, pattern))
)
if(nrow(hits) > 0){
for(n in 1:nrow(hits)){
hit <- hits[n,]
if(stringr::str_detect(tolower(hit$nci_cd_name),
"(mab|art)$|monoclonal antibody|^anti-")){
hit$drug_type <- "Antibody"
}else{
hit$drug_type <- "Small molecule"
}
hit$drug_action_type <- "INHIBITOR"
if(stringr::str_detect(
tolower(hit$nci_cd_name),
"antagonist")){
hit$drug_action_type <- "ANTAGONIST"
}
if(stringr::str_detect(
tolower(hit$nci_cd_name),
"blocker")){
hit$drug_action_type <- "BLOCKER"
}
hit$target_symbol <- target_symbol
hit$target_genename <- target_genename
hit$target_type <- target_type
hit$target_entrezgene <- target_entrezgene
hit$target_ensembl_gene_id <- target_ensembl_gene_id
hit$drug_clinical_source <- "nci_thesaurus_custom"
hit$drug_cancer_relevance <- "by_cancer_target_nci"
hit$drug_n_indications <- 0
hit$drug_frac_cancer_indications <- 0
hit$drug_approved_noncancer <- FALSE
## set general indications for unknown cases
if(is.na(hit$disease_efo_id) &
is.na(hit$disease_efo_label) &
is.na(hit$cui) &
is.na(hit$cui_name)){
hit$disease_efo_id = "EFO:0000311"
hit$disease_efo_label = "cancer"
hit$cui = "C0006826"
hit$cui_name = "Malignant neoplastic disease"
}
custom_nci_targeted_drugs <- custom_nci_targeted_drugs |>
dplyr::bind_rows(hit)
}
}
}
### CHECK HOW MANY TARGET-LACKING INHIBITORS ARE MISSING
### FROM THE CUSTOM NCI MATCHING ROUTINE
inhibitors_no_target_nonmapped <- all_inhibitors_no_target |>
dplyr::anti_join(custom_nci_targeted_drugs, by = "nci_cd_name") |>
dplyr::filter(!stringr::str_detect(
nci_concept_definition,
"(A|a)ntibody(-| )drug conjugate \\(ADC\\)"
)) |>
dplyr::filter(!stringr::str_detect(
tolower(nci_cd_name),"^(allogeneic|regimen |copper |fluorine f |indium |iodine |carbon c|autologous |recombinant |lutetium |yttrium |y 90)|vaccine$"
)) |>
dplyr::filter(
stringr::str_detect(
nci_concept_definition, "antineoplastic|tumor|cancer"
)
) |>
dplyr::select(nci_cd_name,
nci_concept_definition) |>
dplyr::distinct()
drug_df$target_entrezgene <-
as.integer(drug_df$target_entrezgene)
custom_nci_targeted_drugs$target_entrezgene <-
as.integer(custom_nci_targeted_drugs$target_entrezgene)
ot_nci_drugs_curated <-
dplyr::anti_join(drug_df, custom_nci_targeted_drugs,
by = "nci_cd_name") |>
dplyr::bind_rows(custom_nci_targeted_drugs) |>
dplyr::arrange(target_symbol, nci_cd_name) |>
dplyr::mutate(drug_action_type = dplyr::if_else(
(stringr::str_detect(tolower(nci_cd_name),"inhibitor") &
is.na(drug_action_type)) |
(!is.na(nci_cd_name) &
stringr::str_detect(nci_cd_name,"(mab|art)$") &
is.na(drug_action_type)),
"INHIBITOR",
as.character(drug_action_type))) |>
dplyr::mutate(drug_cancer_relevance = dplyr::if_else(
is.na(drug_cancer_relevance) &
(stringr::str_detect(
tolower(nci_concept_definition),
"anti-tumor|chemotherapy|cancer vaccine|immunothera|monoclonal antibody|antineoplastic|treatment of cancer|treatment of metastat") |
stringr::str_detect(tolower(nci_cd_name)," regimen|recombinant|carcinoma|immune checkpoint|anti-programmed cell death ")),
as.character("by_cancer_definition_nci"),
as.character(drug_cancer_relevance)
)) |>
dplyr::mutate(drug_action_type = dplyr::if_else(
is.na(drug_action_type) &
stringr::str_detect(
drug_action_type,
"^(SUBSTRATE|HYDROLYTIC ENZYME|RELEASING AGENT)"),
paste0(drug_action_type,"_OTHER"),
as.character(drug_action_type)
))
duplicated_drugs <- ot_nci_drugs_curated |>
dplyr::group_by(nci_cd_name) |>
dplyr::summarise(drug_cancer_relevance = paste(
sort(unique(drug_cancer_relevance)), collapse="@")) |>
dplyr::filter(stringr::str_detect(drug_cancer_relevance,"@")) |>
tidyr::separate_rows(drug_cancer_relevance, sep = "@") |>
dplyr::filter(!stringr::str_detect(drug_cancer_relevance,"otp"))
ot_nci_drugs_curated <- ot_nci_drugs_curated |>
dplyr::anti_join(duplicated_drugs,
by = c("nci_cd_name","drug_cancer_relevance"))
return(list('curated' = ot_nci_drugs_curated,
'nonmapped' = inhibitors_no_target_nonmapped))
}
assign_drug_category <- function(drug_df = NULL,
path_data_raw = NULL){
atc_classification <-
get_atc_drug_classification(path_data_raw = path_data_raw)
atc_classification_with_drugs <- atc_classification |>
dplyr::mutate(atc_drug_entry = dplyr::case_when(
!is.na(atc_drug_entry) &
(tolower(atc_drug_entry) == "sotorasib" |
tolower(atc_drug_entry) == "ivosidenib" |
tolower(atc_drug_entry) == "enasidenib" |
tolower(atc_drug_entry) == "venetoclax" |
tolower(atc_drug_entry) == "adagrasib") ~ as.character(NA),
TRUE ~ as.character(atc_drug_entry)
)) |>
dplyr::filter(!is.na(atc_drug_entry))
atc_classification_clean <- atc_classification |>
dplyr::filter(is.na(atc_drug_entry)) |>
dplyr::bind_rows(atc_classification_with_drugs) |>
dplyr::select(-atc_drug_entry) |>
dplyr::distinct() |>
dplyr::filter(!is.na(atc_code_level3))
classified_drugs <- list()
classified_drugs[['pre_classified_atc']] <- drug_df |>
dplyr::mutate(drug_entry = tolower(nci_cd_name)) |>
dplyr::left_join(
dplyr::select(
atc_classification_with_drugs,
atc_drug_entry,
atc_code_level3),
by = c("drug_entry" = "atc_drug_entry"),
relationship = "many-to-many") |>
dplyr::filter(!is.na(atc_code_level3)) |>
dplyr::distinct()
drugs_non_classified <- drug_df |>
dplyr::mutate(drug_entry = tolower(nci_cd_name)) |>
dplyr::anti_join(
classified_drugs[['pre_classified_atc']],
by = "drug_entry") |>
dplyr::distinct()
custom_target_classifications <- drugs_non_classified |>
dplyr::filter(!is.na(target_symbol)) |>
dplyr::group_by(drug_name, nci_cd_name,
drug_entry,
nci_concept_definition,
molecule_chembl_id, drug_action_type) |>
dplyr::reframe(
target_symbol = paste(
sort(unique(target_symbol)), collapse="|")) |>
dplyr::mutate(atc_code_level3 = dplyr::case_when(
target_symbol == "ABL1" | target_symbol == "ABL1|BCR" ~ "L01EA",
target_symbol == "BRAF" ~ "L01EC",
stringr::str_detect(target_symbol,"TOP1(\\|TOP2)?") ~ "L01CE",
stringr::str_detect(
target_symbol, "^((PSM(A|B|C|D)[0-9]{1,})|ADRM1|SEM1)") ~ "L01XG",
stringr::str_detect(
target_symbol, "^((BRAF\\|(KDR|RAF))|(ARAF\\|BRAF))") ~ "L01EXJ",
target_symbol == "PLK1" | target_symbol == "PLK4" ~ "L01EXK",
target_symbol == "EGFR" &
!is.na(drug_name) &
stringr::str_detect(drug_name, "MAB") ~ "L01FE",
target_symbol == "ERBB2" &
!is.na(drug_name) &
stringr::str_detect(drug_name, "MAB") ~ "L01FD",
target_symbol == "ERBB2" &
(is.na(drug_name) |
(!is.na(drug_name) &
!stringr::str_detect(drug_name, "MAB"))) ~ "L01EH",
!is.na(drug_name) &
drug_name == "SIROLIMUS" ~ "L01EG",
target_symbol == "EGFR" &
(is.na(drug_name) |
(!is.na(drug_name) &
!stringr::str_detect(drug_name, "MAB"))) ~ "L01EB",
(!is.na(drug_name) &
stringr::str_detect(drug_name, "ABIVERTINIB")) ~ "L01EB",
target_symbol == "ALK" ~ "L01ED",
(drug_action_type == "INHIBITOR" |
drug_action_type == "ANTAGONIST") &
target_symbol == "CTLA4" ~ "L01FXA",
(drug_action_type == "INHIBITOR" |
drug_action_type == "ANTAGONIST") &
target_symbol == "LAG3" ~ "L01FXB",
(drug_action_type == "INHIBITOR" |
drug_action_type == "ANTAGONIST") &
target_symbol == "TIGIT" ~ "L01FXC",
(drug_action_type == "INHIBITOR" |
drug_action_type == "ANTAGONIST") &
(target_symbol == "CD274|CTLA4" |
target_symbol == "CTLA4|PDCD1") ~ "L01FXD",
(!is.na(drug_name) & drug_name == "BRENTUXIMAB VEDOTIN") |
(stringr::str_detect(
target_symbol,"^(TUBA|TUBB)") &
(drug_action_type == "INHIBITOR" |
drug_action_type == "DISRUPTING AGENT")) ~ "L01XXK",
stringr::str_detect(
target_symbol, "^(PARP[1-9]{1}(\\|)?){1,}$") ~ "L01XK",
stringr::str_detect(
target_symbol, "^(HDAC[0-9]{1,}(\\|)?){1,}$") ~ "L01XH",
stringr::str_detect(
target_symbol, "^MAP2K[0-9]") ~ "L01EE",
stringr::str_detect(
target_symbol, "^CDK[0-9]{1,}") ~ "L01EF",
stringr::str_detect(
target_symbol, "^(BIRC|XIAP)") ~ "L01XXF",
stringr::str_detect(
target_symbol, "^JAK[1-2]") ~ "L01EJ",
stringr::str_detect(
target_symbol, "^AURK(A|B|C)") ~ "L01EXB",
stringr::str_detect(
target_symbol, "EGFR") &
stringr::str_detect(
target_symbol,"ERBB") ~ "L01EXH",
stringr::str_detect(
target_symbol, "^FGFR[1-4]{1}") ~ "L01EN",
stringr::str_detect(
target_symbol, "^IDH[1-2]{1}") ~ "L01XXD",
stringr::str_detect(
target_symbol, "^((IGF1\\|)|IGF1R)") ~ "L01XXN",
!is.na(drug_name) &
stringr::str_detect(
tolower(drug_name), "^aberaterone") ~ "L02BX",
!is.na(drug_name) &
stringr::str_detect(
tolower(drug_name), "^(gemcitabine)") ~ "L01BC",
!is.na(drug_name) &
stringr::str_detect(
tolower(drug_name), "^(fludarabine)") ~ "L01BB",
stringr::str_detect(
target_symbol, "^(K|N|H)RAS") ~ "L01XXC",
stringr::str_detect(
target_symbol, "^MET$") ~ "L01EXA",
stringr::str_detect(
target_symbol, "^(CD3D\\|CD3E\\|CD3G)") ~ "L01FXE",
stringr::str_detect(
target_symbol, "^(AKT[0-9](\\|)?){1,}") |
(!is.na(drug_name) &
stringr::str_detect(
drug_name, "GSK-690693")) ~ "L01EXC",
stringr::str_detect(
target_symbol, "^ATR$") ~ "L01XXI",
stringr::str_detect(
target_symbol, "^PDGFR(A|B)|^KIT") ~ "L01EXE",
stringr::str_detect(
target_symbol, "^(NTRK[0-9]\\|?){1,}$") |
(!is.na(drug_name) & drug_name == "TALETRECTINIB") ~ "L01EXF",
stringr::str_detect(
target_symbol, "^(CHEK(1|2)\\|?){1,2}$") ~ "L01XXJ",
stringr::str_detect(
target_symbol, "^BCL2") ~ "L01XXM",
stringr::str_detect(
target_symbol, "^GNRH") ~ "L02AE",
stringr::str_detect(
target_symbol, "^BRD(T|[1-9]{1})") ~ "L01XXA",
stringr::str_detect(
target_symbol, "^MTOR\\|PIK3") ~ "L01XXG",
stringr::str_detect(
target_symbol, "^PIK3") |
(!is.na(drug_name) &
(stringr::str_detect(drug_name,"UMBRALISIB") |
(drug_name == "OMIPALISIB" |
drug_name == "PF-04691502"))) ~ "L01EM",
stringr::str_detect(
target_symbol, "^BTK$") ~ "L01EL",
(!is.na(drug_name) &
(drug_name == "PRALSETINIB" |
drug_name == "SELPERCATINIB")) ~ "L01EXL",
stringr::str_detect(
target_symbol, "^MTOR$") ~ "L01EG",
stringr::str_detect(
target_symbol, "^(KDR|FLT1|FLT3|FLT4)") ~ "L01EK",
stringr::str_detect(
target_symbol, "^(MS4A1)") ~ "L01FA",
stringr::str_detect(
target_symbol, "^(ESR1)") ~ "L02BA",
stringr::str_detect(
target_symbol, "^(CD38)") ~ "L01FC",
stringr::str_detect(
target_symbol, "^(CD22)") ~ "L01FB",
target_symbol == "MAPK1|MAPK3" ~ "L01XXE",
stringr::str_detect(target_symbol, "^AR$") &
(!is.na(nci_concept_definition) &
stringr::str_detect(
nci_concept_definition, "androgen receptor")) ~ "L02BB",
(target_symbol == "PDCD1" |
target_symbol == "CD274") ~ "L01FF",
TRUE ~ as.character(NA)
)) |>
dplyr::mutate(atc_code_level3 = dplyr::if_else(
(!is.na(drug_name) &
stringr::str_detect(
drug_name,
paste0(
"DOVITINIB|BRIVANIB|UCN-01|AT-9283|",
"SURUFATINIB|ORANTINIB|LUCITANIB|",
"ALTIRATINIB|CEP-11981|CRENOLANIB|",
"FORETINIB|FAMITINIB|MOTESANIB|",
"LESTAURTINIB|OSI-930|PACRITINIB|KW-2449|",
"CABOZANTINIB|BMS-817378|BMS-794833|",
"GOLVATINIB|TAK-593|XL-820|TANDUTINIB|IMATINIB|",
"CERDULATINIB|CEP-2563|FEDRATINIB|IBCASERTIB|",
"VANDETANIB|TESEVATINIB|TARLOXOTINIB|",
"TAK-285|SKLB1028|PUQUITINIB|KBP5209|REPOTRECTINIB|",
"VATALANIB|QUIZARTINIB|BMS-690514|IMATINIB|DASATINIB|",
"CANERTINIB|CEP-32496|REGORAFENIB|GUSACITINIB|",
"SU-014813|X-82|XL-999|LINIFANIB|NINGETINIB|",
"PEXIDARTINIB|RG-1530|SITRAVATINIB|SORAFENIB|SUNITINIB|",
"ENTRECTINIB"))) |
(!is.na(nci_concept_definition) &
stringr::str_detect(
tolower(nci_concept_definition),
paste0(
"multikinase|multi(-)?targeted|multi-kinase inhibitor|",
"targets multiple|inhibitor of multiple|multiple-receptor")) &
stringr::str_count(target_symbol, pattern = "\\|") >= 2),
"L01EXI",
as.character(atc_code_level3)
)) |>
dplyr::left_join(
dplyr::select(
atc_classification_clean,
atc_code_level3,
),
by = "atc_code_level3",
relationship = "many-to-many"
) |>
#dplyr::filter(!is.na(atc_code_level3)) |>
dplyr::select(
#molecule_chembl_id,
drug_entry,
atc_code_level3
) |>
dplyr::group_by(drug_entry) |>
dplyr::summarise(
atc_code_level3 = paste(
unique(atc_code_level3), collapse="|")) |>
dplyr::distinct() |>
dplyr::mutate(
atc_code_level3 = stringr::str_replace(
atc_code_level3, "^(NA\\|)|(\\|(NA|L01XXK))$",""
)
)
classified_drugs[['classified_targeted_custom']] <-
drugs_non_classified |>
dplyr::filter(!is.na(target_symbol)) |>
dplyr::left_join(
custom_target_classifications, by = "drug_entry",
relationship = "many-to-many") |>
dplyr::distinct()
classified_drugs_all<-
classified_drugs[['pre_classified_atc']] |>
dplyr::bind_rows(
classified_drugs[['classified_targeted_custom']]) |>
dplyr::bind_rows(
dplyr::filter(drugs_non_classified, is.na(target_symbol))) |>
dplyr::mutate(atc_code_level3 = dplyr::case_when(
(is.na(atc_code_level3) | atc_code_level3 == "NA") &
(!is.na(nci_concept_definition) &
stringr::str_detect(
tolower(nci_concept_definition),
"anthracycline|anthracenedione")) ~ "L01DB",
(is.na(atc_code_level3) | atc_code_level3 == "NA") &
stringr::str_detect(tolower(drug_entry),"hydroxyurea|leucovorin|eniluracil") ~ "L01BA",
(is.na(atc_code_level3) | atc_code_level3 == "NA") &
stringr::str_detect(tolower(drug_entry),"arsenic trioxide") ~ "L01BA",
(is.na(atc_code_level3) | atc_code_level3 == "NA") &
stringr::str_detect(tolower(drug_entry),
"thioguanine|aspacytarabine|aspacytarabine|troxacitabine") ~ "L01BB",
(is.na(atc_code_level3) | atc_code_level3 == "NA") &
!stringr::str_detect(drug_entry,"/| ") &
stringr::str_detect(drug_entry, "xel$") ~ "L01CD",
(is.na(atc_code_level3) | atc_code_level3 == "NA") &
!stringr::str_detect(drug_entry,"/| ") &
stringr::str_detect(drug_entry, "platin$") ~ "L01XA",
(is.na(atc_code_level3) | atc_code_level3 == "NA") &
!is.na(nci_concept_definition) &
stringr::str_detect(tolower(nci_concept_definition), "anti-estrogen") ~ "L02BA",
(is.na(atc_code_level3) | atc_code_level3 == "NA") &
!is.na(nci_concept_definition) &
stringr::str_detect(tolower(nci_concept_definition), "aromatase inhibitor") ~ "L02BG",
(is.na(atc_code_level3) | atc_code_level3 == "NA") &
!is.na(nci_concept_definition) &
stringr::str_detect(tolower(nci_concept_definition), "nitrogen mustard") ~ "L01AA",
(is.na(atc_code_level3) | atc_code_level3 == "NA") &
stringr::str_detect(
tolower(nci_concept_definition), "purine( nucleoside)? analog") ~ "L01BB",
(is.na(atc_code_level3) | atc_code_level3 == "NA") &
stringr::str_detect(
tolower(nci_concept_definition), "pyrimidine( nucleoside)? analog") ~ "L01BC",
(is.na(atc_code_level3) | atc_code_level3 == "NA") &
!is.na(drug_entry) &
!stringr::str_detect(drug_entry, "/") &
stringr::str_detect(
tolower(nci_concept_definition), "vinca alkaloid") ~ "L01CA",
(is.na(atc_code_level3) | atc_code_level3 == "NA") &
((!is.na(nci_concept_definition) &
stringr::str_detect(
tolower(nci_concept_definition),
"antineoplastic activit|anti-tumor activit"
)) |
(!is.na(drug_max_ct_phase) &
stringr::str_detect(
drug_entry,"(in|ib|ide|ine|ax|il|an|ate| alfa)$") &
drug_max_ct_phase >= 2 &
(!is.na(drug_n_indications) &
drug_n_indications > 2) &
(!is.na(drug_frac_cancer_indications) &
drug_frac_cancer_indications > 0.4))) ~ "L01XX",
TRUE ~ as.character(atc_code_level3)
)) |>
dplyr::distinct() |>
dplyr::group_by(dplyr::across(-c("atc_code_level3"))) |>
dplyr::summarise(atc_code_level3 = paste(unique(atc_code_level3), collapse="|"),
.groups = "drop") |>
dplyr::mutate(atc_code_level3 = dplyr::case_when(
stringr::str_detect(
atc_code_level3, "^(L0[A-Z0-9]{1,}\\|((S|M)01(XA|LA|AH)))$") ~
stringr::str_replace_all(
atc_code_level3, "(\\|((S|M)01(XA|LA|AH)))$", ""),
stringr::str_detect(
atc_code_level3, "L0[A-Z0-9]{1,}\\|L01XX") ~
stringr::str_replace_all(
atc_code_level3, "\\|L01XX", ""),
atc_code_level3 == "D11AH|L01EJ" |
atc_code_level3 == "D11AH|L01XF" |
atc_code_level3 == "G03DA|L02AB" |
atc_code_level3 == "D10AD|L01XF" |
atc_code_level3 == "D11AH|L04AD" ~
stringr::str_replace_all(
atc_code_level3, "^(D10AD|G03DA|D11AH)\\|", ""),
atc_code_level3 == "L01FG|S01LA" |
atc_code_level3 == "L01EG|S01XA" |
atc_code_level3 == "L01BB|L04AA" |
atc_code_level3 == "L01BA|L04AX" |
atc_code_level3 == "L01XX|M01AH" ~
stringr::str_replace_all(
atc_code_level3, "\\|(S01LA|S01XA|L04AA|L04AX|M01AH)", ""),
TRUE ~ atc_code_level3
)) |>
dplyr::distinct()
atc_classified_drugs <- classified_drugs_all |>
dplyr::filter(!is.na(drug_entry)) |>
dplyr::select(atc_code_level3, drug_entry) |>
dplyr::filter(!is.na(atc_code_level3)) |>
tidyr::separate_rows(atc_code_level3, sep = "\\|") |>
dplyr::filter(!is.na(atc_code_level3)) |>
dplyr::left_join(
dplyr::select(
atc_classification_clean,
atc_code_level1,
atc_level1,
atc_code_level2,
atc_level2,
atc_code_level3,
atc_level3
), relationship ="many-to-many"
) |>
dplyr::group_by(drug_entry) |>
dplyr::summarise(
atc_code_level1 = paste(unique(atc_code_level1), collapse="|"),
atc_level1 = paste(unique(stringr::str_trim(atc_level1)), collapse="|"),
atc_code_level2 = paste(unique(atc_code_level2), collapse="|"),
atc_level2 = paste(unique(stringr::str_trim(atc_level2)), collapse="|"),
atc_code_level3 = paste(unique(atc_code_level3), collapse="|"),
atc_level3 = paste(unique(stringr::str_trim(atc_level3)), collapse="|")) |>
dplyr::mutate(atc_code_level1 = dplyr::if_else(
!is.na(atc_code_level1),
stringr::str_replace(
atc_code_level1,"(^NA\\|)|\\|NA$",""),
as.character(atc_code_level1)
)) |>
dplyr::mutate(atc_level1 = dplyr::if_else(
!is.na(atc_level1),
stringr::str_replace(
atc_level1,"(^NA\\|)|\\|NA$",""),
as.character(atc_level1)
)) |>
dplyr::mutate(atc_code_level2 = dplyr::if_else(
!is.na(atc_code_level2),
stringr::str_replace(
atc_code_level2,"(^NA\\|)|\\|NA$",""),
as.character(atc_code_level2)
)) |>
dplyr::mutate(atc_level2 = dplyr::if_else(
!is.na(atc_level2),
stringr::str_replace(
atc_level2,"(^NA\\|)|\\|NA$",""),
as.character(atc_level2)
)) |>
dplyr::mutate(atc_code_level3 = dplyr::if_else(
!is.na(atc_code_level3),
stringr::str_replace(
atc_code_level3,"(^NA\\|)|\\|NA$",""),
as.character(atc_code_level3)
)) |>
dplyr::mutate(atc_level3 = dplyr::if_else(
!is.na(atc_level3),
stringr::str_replace(
atc_level3,"(^NA\\|)|\\|NA$",""),
as.character(atc_level3)
)) |>
dplyr::mutate(atc_code_level3 = dplyr::if_else(
!is.na(atc_code_level3),
stringr::str_replace(
atc_code_level3,"L01XX\\|L01XXK","L01XXK"),
as.character(atc_code_level3)
)) |>
dplyr::mutate(atc_level3 = dplyr::if_else(
!is.na(atc_level3),
stringr::str_replace(
atc_level3,
"Other antineoplastic agents\\|Tubulin inhibitors",
"Tubulin inhibitors"),
as.character(atc_level3)
)) |>
dplyr::mutate(atc_level3 = dplyr::if_else(
!is.na(atc_level3),
stringr::str_replace(
atc_level3,
"Folic acid analogues|Other immunosuppressants",
"Folic acid analogues"),
as.character(atc_level3)
))
classified_drugs_all_final <- classified_drugs_all |>
dplyr::select(-atc_code_level3) |>
dplyr::left_join(atc_classified_drugs, by = "drug_entry",
relationship = "many-to-many") |>
dplyr::mutate(atc_treatment_category = "unknown") |>
dplyr::mutate(atc_treatment_category = dplyr::case_when(
!is.na(atc_code_level2) &
stringr::str_detect(
atc_code_level2, "(\\|L02(A|B))|(^L02(A|B))") ~
"cancer_hormone_therapy",
!is.na(target_symbol) &
!is.na(atc_code_level2) &
stringr::str_detect(
atc_code_level2, "(\\|L01(E|F))|(^L01(E|F))"
) ~ "cancer_targeted_therapy",
!is.na(target_symbol) &
!is.na(atc_code_level3) &
stringr::str_detect(
atc_code_level3, "(\\|L01XX[A-J])|(^L01XX[A-J])"
) ~ "cancer_targeted_therapy",
!is.na(target_symbol) &
!is.na(atc_code_level3) &
stringr::str_detect(
atc_code_level3, "(\\|L01XXM)|(^L01XXM)"
) ~ "cancer_targeted_therapy",
!is.na(target_symbol) &
!is.na(atc_code_level3) &
stringr::str_detect(
atc_code_level3, "(\\|L04)|(^L04)"
) ~ "cancer_immuno_suppressants",
!is.na(target_symbol) &
stringr::str_detect(drug_cancer_relevance,"^by_cancer") &
(is.na(atc_code_level3) |
atc_code_level3 == "NA" |
atc_code_level3 == "L01XX")
~ "cancer_unclassified",
!is.na(target_symbol) &
!is.na(atc_code_level3) &
stringr::str_detect(
atc_code_level3, "(\\|L01X[A-L])|(^L01X[A-L])"
) ~ "cancer_targeted_therapy",
!is.na(atc_code_level2) &
!stringr::str_detect(
atc_code_level2, "(\\|L01(E|F))|(^L01(E|F))") &
((!is.na(atc_code_level3) &
drug_type != "Antibody" &
stringr::str_detect(atc_code_level3, "L01XXK")) |
(!is.na(atc_code_level3) &
stringr::str_detect(atc_code_level3,"L01X(A|B)")) |
(stringr::str_detect(
atc_code_level2, "(L01(A|B|C|D))"
))) ~ "cancer_chemo_therapy",
(!is.na(atc_code_level3) &
drug_type == "Antibody" &
stringr::str_detect(
atc_code_level3, "L01XXK")
) ~ "cancer_adc_targeted_therapy",
!is.na(target_symbol) &
!is.na(atc_code_level2) &
atc_code_level2 != "NA" &
!stringr::str_detect(
atc_code_level2, "\\|L0|^L0"
) ~ "other_targeted_therapy",
TRUE ~ as.character(atc_treatment_category)
))
cancer_targets_classified <- classified_drugs_all_final |>
dplyr::filter(
(drug_cancer_relevance == "by_cancer_condition_otp" |
drug_cancer_relevance == "by_cancer_target_otp" |
drug_cancer_relevance == "by_cancer_target_nci") &
atc_treatment_category == "cancer_targeted_therapy") |>
dplyr::select(target_symbol) |>
dplyr::filter(!is.na(target_symbol)) |>
dplyr::filter(!stringr::str_detect(target_symbol,"^(RP(S|L)|TUB)")) |>
dplyr::distinct() |>
dplyr::mutate(cancer_target_classified = TRUE)
classified_drugs_all_final <- classified_drugs_all_final |>
dplyr::left_join(cancer_targets_classified) |>
dplyr::mutate(cancer_target_classified = dplyr::if_else(
is.na(target_symbol) | is.na(cancer_target_classified),
FALSE,
as.logical(cancer_target_classified)
))
drug_df <- remove_duplicate_chembl_ids(
drug_df = classified_drugs_all_final)
drug_df$drug_entry <- NULL
return(drug_df)
}
remove_duplicate_chembl_ids <- function(drug_df = NULL){
black_list <-
readr::read_tsv(
file = "data-raw/drug_name_black_list.txt",
col_names = F, show_col_types = F)
black_list$nci_cd_name <- black_list$X1
drug_df <- drug_df |> dplyr::anti_join(
black_list, by = "nci_cd_name"
)
return(drug_df)
}
clean_final_drug_list <- function(drug_df = NULL){
pharmaoncox <- drug_df |>
dplyr::filter(
is.na(nci_cd_name) |
!stringr::str_detect(
nci_cd_name,
"Sustained-release| Bead(s)?| Compound|Vector|Pegylated")
) |>
dplyr::filter(
is.na(nci_concept_definition) |
!stringr::str_detect(
tolower(nci_concept_definition),
"a( (synthetic|diagnostic|targeted))?( radio(immuno)conjugate)")
) |>
dplyr::filter(
is.na(drug_action_type) |
!stringr::str_detect(tolower(drug_action_type),"vaccine")
) |>
dplyr::filter(
is.na(nci_concept_definition) |
!stringr::str_detect(
tolower(nci_concept_definition),
"(^(a|any)|cancer|tumor|dna|autologous|cell-based|synthetic|cell|peptide|valent)( cell)? vaccine")
) |>
dplyr::filter(
is.na(nci_cd_name) |
!stringr::str_detect(
nci_cd_name,
"^(Carbon C|Fluorine F|Gallium Ga|Indium In|Iodine I|Lutetium Lu|Technetium Tc|Yttrium Y)"
)
) |>
dplyr::filter(is.na(nci_cd_name) |
nci_cd_name != "Abivertinib Maleate") |>
dplyr::mutate(nci_cd_name = dplyr::if_else(
is.na(nci_cd_name),
stringr::str_to_title(drug_name),
as.character(nci_cd_name)
)) |>
dplyr::mutate(is_salt = dplyr::if_else(
nci_cd_name == "Avitinib Maleate",
as.logical(NA),
as.logical(is_salt)
)) |>
dplyr::mutate(drug_name = dplyr::if_else(
nci_cd_name == "Doxycycline" &
!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL1433",
"DOXYCYCLINE",
as.character(drug_name)))
drug_action_types <- as.data.frame(
pharmaoncox |>
dplyr::select(nci_cd_name, drug_action_type) |>
dplyr::distinct() |>
dplyr::group_by(nci_cd_name) |>
dplyr::summarise(drug_action_type = paste(
drug_action_type, collapse = "/"
))
)
pharmaoncox$drug_action_type <- NULL
pharmaoncox <- pharmaoncox |>
dplyr::left_join(
drug_action_types,
by = "nci_cd_name",
relationship = "many-to-many") |>
dplyr::select(drug_name,
nci_cd_name,
drug_type,
drug_action_type,
molecule_chembl_id,
drug_moa,
drug_max_phase_indication,
dplyr::everything())
drug_max_ct_phase <- as.data.frame(
pharmaoncox |>
dplyr::select(nci_cd_name, drug_max_ct_phase) |>
dplyr::group_by(nci_cd_name) |>
dplyr::summarise(drug_max_ct_phase = max(drug_max_ct_phase))
)
pharmaoncox$drug_max_ct_phase <- NULL
pharmaoncox <- pharmaoncox |>
dplyr::left_join(drug_max_ct_phase,
by = "nci_cd_name",
relationship = "many-to-many") |>
dplyr::select(-c(drug_moa)) |>
dplyr::mutate(nci_concept_synonym2 = dplyr::if_else(
is.na(nci_concept_synonym_all) & !is.na(drug_synonyms),
as.character(tolower(drug_synonyms)),
as.character(tolower(nci_concept_synonym_all))
)) |>
dplyr::mutate(nci_concept_synonym_all2 = nci_concept_synonym_all) |>
tidyr::separate_rows(nci_concept_synonym2,
sep="\\|") |>
dplyr::rename(nci_concept_synonym = nci_concept_synonym2) |>
dplyr::select(-c(nci_concept_synonym_all2,
drug_synonyms,
drug_tradenames,
drug_description)) |>
dplyr::distinct() |>
dplyr::select(drug_name, nci_cd_name, drug_type,
drug_action_type, drug_cancer_relevance,
molecule_chembl_id,
drug_max_phase_indication, drug_max_ct_phase,
target_genename, target_symbol,
target_type, target_ensembl_gene_id,
target_entrezgene,
disease_efo_id, disease_efo_label,
cui, cui_name, primary_site,
nci_concept_synonym,
nci_concept_synonym_all,
dplyr::everything()) |>
dplyr::mutate(nci_concept_definition =
stringi::stri_enc_toascii(nci_concept_definition)) |>
dplyr::mutate(nci_concept_synonym_all =
stringi::stri_enc_toascii(nci_concept_synonym_all)) |>
dplyr::mutate(nci_concept_synonym =
stringi::stri_enc_toascii(nci_concept_synonym)) |>
dplyr::mutate(drug_name =
stringi::stri_enc_toascii(drug_name)) |>
dplyr::mutate(
nci_cd_name =
stringi::stri_enc_toascii(nci_cd_name)) |>
dplyr::mutate(drug_action_type = stringr::str_replace_all(
drug_action_type, "/NA|NA/",""
)) |>
dplyr::mutate(idx = dplyr::row_number())
## Simplify records with only "cancer" indications,
## mapping them to a unique
## EFO/CUI cross-ref, avoiding similar records
##with "neoplasm", "carcinoma" etc.
pharmaoncox_cancer_no_indication <- pharmaoncox |>
dplyr::filter(is.na(disease_efo_id))
pharmaoncox_non_cancer <- pharmaoncox |>
dplyr::filter(is.na(primary_site) & !is.na(disease_efo_id) &
drug_cancer_relevance == "by_other_condition_otp")
pharmaoncox_cancer_NOS <- as.data.frame(
pharmaoncox |>
dplyr::filter(is.na(primary_site) & !is.na(disease_efo_id) &
drug_cancer_relevance != "by_other_condition_otp") |>
dplyr::mutate(disease_efo_id = "EFO:0000311",
disease_efo_label = "cancer",
cui = "C0006826",
cui_name = "Malignant neoplastic disease") |>
dplyr::group_by(
dplyr::across(-dplyr::ends_with(c("drug_clinical_id")))) |>
dplyr::summarise(
drug_clinical_id = paste(unique(drug_clinical_id), collapse=","),
.groups = "drop"
) |>
dplyr::distinct()
)
pharmaoncox_cancer_specific <- pharmaoncox |>
dplyr::filter(!is.na(primary_site))
pharmaoncox <- pharmaoncox_cancer_no_indication |>
dplyr::bind_rows(pharmaoncox_non_cancer) |>
dplyr::bind_rows(pharmaoncox_cancer_specific) |>
dplyr::bind_rows(pharmaoncox_cancer_NOS) |>
dplyr::arrange(nci_cd_name) |>
dplyr::mutate(nci_cd_name = dplyr::if_else(
stringr::str_detect(nci_cd_name,";-"),
drug_name,
as.character(nci_cd_name)
)) |>
dplyr::filter(!is.na(nci_cd_name)) |>
dplyr::mutate(drug_name_final = dplyr::case_when(
!is.na(drug_name) &
!is.na(nci_cd_name) &
tolower(nci_cd_name) != tolower(drug_name) &
stringr::str_detect(
tolower(nci_cd_name),"(ib|mab|in)$") &
stringr::str_detect(
drug_name, "[0-9]{1,}") ~ nci_cd_name,
!is.na(drug_name) &
!is.na(nci_cd_name) &
tolower(nci_cd_name) != tolower(drug_name) &
!stringr::str_detect(
tolower(nci_cd_name),"(ib|mab|in)$") &
stringr::str_detect(
drug_name, "[0-9]{1,}") ~ drug_name,
!is.na(drug_name) &
!is.na(nci_cd_name) &
tolower(nci_cd_name) == tolower(drug_name) ~
stringr::str_to_title(tolower(nci_cd_name)),
is.na(drug_name) & !is.na(nci_cd_name) ~ nci_cd_name,
TRUE ~ as.character(
stringr::str_to_title(tolower(drug_name))
)
)) |>
dplyr::filter(
!(drug_name_final == "CFI-400945" &
molecule_chembl_id == "CHEMBL3408947") &
!(drug_name_final == "Relebactam" &
molecule_chembl_id == "CHEMBL3112741") &
!(drug_name_final == "SR16234" &
molecule_chembl_id == "CHEMBL3545210")
)
pharmaoncox$drug_name <- NULL
pharmaoncox$drug_name <- pharmaoncox$drug_name_final
pharmaoncox$nci_cd_name <- NULL
pharmaoncox$drug_name_final <- NULL
pharmaoncox$nci_concept_synonym <- NULL
blackbox_warnings <- as.data.frame(pharmaoncox |>
dplyr::select(drug_name, drug_blackbox_warning) |>
dplyr::distinct() |>
dplyr::group_by(drug_name) |>
dplyr::summarise(drug_blackbox_warning = paste(
unique(drug_blackbox_warning), collapse=","))
)
nci_t_map <- as.data.frame(pharmaoncox |>
dplyr::select(drug_name, nci_t, nci_concept_definition) |>
dplyr::distinct() |>
dplyr::group_by(drug_name) |>
dplyr::summarise(
nci_t = paste(
unique(nci_t), collapse=","),
nci_concept_definition = paste(
unique(nci_concept_definition),
collapse = ",")
), .groups = "drop"
)
pharmaoncox$nci_t <- NULL
pharmaoncox$nci_concept_definition <- NULL
pharmaoncox$drug_blackbox_warning <- NULL
pharmaoncox2 <- pharmaoncox |>
dplyr::left_join(
nci_t_map, by = "drug_name",
relationship = "many-to-many") |>
dplyr::left_join(
blackbox_warnings, by = "drug_name",
relationship = "many-to-many") |>
dplyr::distinct() |>
dplyr::mutate(drug_action_type = dplyr::if_else(
drug_action_type == "NA" &
stringr::str_detect(tolower(drug_name),
"ib |ib$") &
stringr::str_detect(tolower(nci_concept_definition),
"inhibitor"),
"INHIBITOR",
as.character(drug_action_type)
)) |>
dplyr::mutate(drug_type = dplyr::if_else(
drug_action_type == "NA" &
stringr::str_detect(tolower(drug_name),
"ib |ib$") &
stringr::str_detect(tolower(nci_concept_definition),
"inhibitor"),
"Small molecule",
as.character(drug_type)
)) |>
dplyr::mutate(inhibition_moa = dplyr::if_else(
!stringr::str_detect(
drug_action_type,
paste0(
"^(POSITIVE|NEGATIVE|AGONIST|HYDROLYTIC|",
"CHELATING|PARTIAL|NA|INVERSE|OPENER|",
"RELEASING|ANTISENSE|STABILISER|MODULATOR|",
"SEQUESTERING|OXIDATIVE|OTHER|ACTIVATOR|RNA|",
"VACCINE|DEGRADER|PROTEOLYTIC|SUBSTRATE)")) &
(is.na(drug_type) |
(drug_type != "Protein" &
drug_type != "Gene" &
drug_type != "Cell")) &
(is.na(target_type) |
target_type != "protein_protein_interaction"),
as.logical(TRUE),
as.logical(FALSE),
as.logical(FALSE)
)) |>
dplyr::mutate(inhibition_moa = dplyr::if_else(
inhibition_moa == F &
drug_action_type == "OTHER" &
(stringr::str_detect(
tolower(drug_name),"mab$") |
drug_type == "Antibody"),
as.logical(TRUE),
as.logical(inhibition_moa)
))
inhibition_moa_df <- as.data.frame(pharmaoncox2 |>
dplyr::select(drug_name, inhibition_moa) |>
dplyr::distinct() |>
dplyr::group_by(drug_name) |>
dplyr::summarise(inhibition_moa = paste(
unique(inhibition_moa), collapse="|"
), .groups = "drop")
) |>
dplyr::mutate(inhibition_moa = dplyr::if_else(
stringr::str_detect(inhibition_moa, "\\|"),
as.logical(TRUE),
as.logical(inhibition_moa)
))
pharmaoncox2$inhibition_moa <- NULL
pharmaoncox2 <- pharmaoncox2 |>
dplyr::left_join(
inhibition_moa_df,
by = "drug_name",
relationship = "many-to-many") |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL1742994") &
drug_name == "Brentuximab vedotin")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL1908394") &
drug_name == "GSK-461364")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL2158685") &
drug_name == "ABC-294640")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL2331680") &
drug_name == "RG-7603")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL253969") &
drug_name == "OSI-632")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL283120") &
drug_name == "AXL-1717")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL52885") &
drug_name == "ENMD-981693")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL4650827") &
drug_name == "SNDX-5613 FREE BASE")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL1645462") &
drug_name == "AC-480")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL4303241") &
drug_name == "BAY-1161909")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL4298098") &
drug_name == "SAR-408701")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL4091801") &
drug_name == "APG115")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL3623290") &
drug_name == "AZD-3759")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL103") &
drug_name == "Mycophenolic Acid")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL3545007") &
drug_name == "RG-7602")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL3545003") &
drug_name == "MSC-2363318A")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL4297310") &
drug_name == "ONC-201")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL4297930") &
drug_name == "CYC-065")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL3899477") &
(nci_t == "NA" | is.na(nci_t)))) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL3126004") &
(nci_t == "NA" | is.na(nci_t)))) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL376408") &
(nci_t == "NA" | is.na(nci_t)))) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL4297489") &
(nci_t == "NA" | is.na(nci_t)))) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL17157") &
(nci_t == "NA" | is.na(nci_t)))) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL1201231") &
(nci_t == "NA" | is.na(nci_t)))) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL1433") &
(nci_t == "NA" | is.na(nci_t)))) |>
dplyr::filter(!(drug_name == "Risedronate Sodium" |
drug_name == "Reminertant" |
drug_name == 'Ibandronate Sodium'))
drug_maps <- list()
drug_maps[['id2name']] <- pharmaoncox2 |>
dplyr::select(drug_name) |>
dplyr::distinct() |>
dplyr::mutate(
drug_id = dplyr::row_number())
if(nrow(drug_maps[['id2name']]) != length(
unique(pharmaoncox2$drug_name))){
lgr::lgr$fatal("Ambiguous drug names")
return(0)
}
pharmaoncox2 <- pharmaoncox2 |>
dplyr::left_join(
drug_maps[['id2name']],
by = "drug_name",
multiple = "all")
drug_maps[['id2target']] <- pharmaoncox2 |>
dplyr::select(drug_id,
target_symbol,
target_entrezgene,
target_genename,
target_ensembl_gene_id,
target_type) |>
dplyr::distinct()
drug_maps[['id2indication']] <- pharmaoncox2 |>
dplyr::select(drug_id,
drug_max_phase_indication,
drug_approved_indication,
drug_frac_cancer_indications,
drug_approved_noncancer,
drug_n_indications,
drug_clinical_source,
drug_year_first_approval,
drug_clinical_id,
drug_max_ct_phase,
disease_efo_id,
disease_efo_label,
cui,
cui_name,
primary_site) |>
dplyr::distinct()
drug_maps[['id2synonym']] <- pharmaoncox2 |>
dplyr::select(drug_id,
nci_concept_synonym_all) |>
dplyr::distinct()
drug_maps[['id2basic']] <- as.data.frame(
pharmaoncox2 |>
dplyr::select(
drug_id,
molecule_chembl_id,
drug_type,
drug_action_type,
drug_cancer_relevance,
inhibition_moa,
is_salt,
is_adc,
drug_blackbox_warning,
nci_concept_definition,
nci_t,
opentargets,
atc_code_level1,
atc_level1,
atc_code_level2,
atc_level2,
atc_code_level3,
atc_level3,
atc_treatment_category) |>
dplyr::distinct())
return(drug_maps)
}
expand_drug_aliases <- function(drug_index_map = NULL,
path_data_raw = NULL,
chembl_pubchem_datestamp = chembl_pubchem_datestamp){
chembl_pubchem_xref <-
get_chembl_pubchem_xref(
datestamp = chembl_pubchem_datestamp,
path_data_raw = path_data_raw)
drugAliasPrimary <-
drug_index_map[['id2name']] |>
dplyr::select(drug_id, drug_name) |>
dplyr::mutate(alias = drug_name) |>
dplyr::select(-drug_name) |>
dplyr::mutate(alias_source = "primaryName")
drugAliasNCI <- as.data.frame(
drug_index_map[['id2synonym']] |>
dplyr::select(drug_id, nci_concept_synonym_all) |>
tidyr::separate_rows(nci_concept_synonym_all, sep="\\|") |>
dplyr::distinct() |>
dplyr::rename(alias = nci_concept_synonym_all) |>
dplyr::filter(nchar(alias) > 3) |>
dplyr::mutate(alias_source = "nci")
)
## Extend aliases with those found in PubChem
## get drug set that contains PubChem cross-references
unique_chembl_pubchem <- drug_index_map[['id2basic']] |>
dplyr::select(drug_id, molecule_chembl_id) |>
dplyr::filter(!is.na(molecule_chembl_id)) |>
dplyr::distinct() |>
dplyr::left_join(
chembl_pubchem_xref,
by = "molecule_chembl_id",
multiple = "all", relationship = "many-to-many") |>
dplyr::filter(!is.na(pubchem_cid)) |>
dplyr::select(-c(chembl_db_version))
## Retrieve aliases for drugs with PubChem x-refs
lgr::lgr$info("Retrieving additional drug aliases/synonyms for PubChem-mapped compounds")
pubchem_synonym_files <-
sort(list.files(path = file.path(path_data_raw, "pubchem"),
pattern = "CID-Synonym-filtered_",
full.names = T))
drugAliasPubchem <- data.frame()
for(f in pubchem_synonym_files){
synonym_data <- as.data.frame(readr::read_tsv(
f, col_names = c('pubchem_cid','alias'),
col_types = "dc",
progress = F
))
pubchem_alias_df <- synonym_data |>
dplyr::inner_join(
unique_chembl_pubchem,
by = "pubchem_cid",
multiple = "all", relationship = "many-to-many")
if(nrow(pubchem_alias_df) > 0){
pubchem_alias_df <- pubchem_alias_df |>
dplyr::select(-c(pubchem_cid, molecule_chembl_id)) |>
dplyr::mutate(alias_source = "pubchem")
drugAliasPubchem <-
drugAliasPubchem |>
dplyr::bind_rows(pubchem_alias_df)
}
rm(synonym_data)
}
## ignore some drugs
custom_name_ignore <- as.data.frame(readr::read_tsv(
file = "data-raw/drug_names_ignore.tsv",
col_names = F, show_col_types = F))
colnames(custom_name_ignore) <- c("alias_lc")
custom_name_ignore$alias_lc <- tolower(
custom_name_ignore$alias_lc)
drugAliasAll <- as.data.frame(drugAliasPrimary |>
dplyr::bind_rows(drugAliasNCI) |>
dplyr::bind_rows(drugAliasPubchem) |>
dplyr::filter(nchar(alias) > 3 &
nchar(alias) < 250) |>
#tidyr::separate_rows(alias, sep = "\\|") |>
dplyr::distinct() |>
dplyr::left_join(drug_index_map[['id2name']],
by = "drug_id") |>
## avoid drug aliases for a particular drug that are identical
## to the primary drug name of another drug
dplyr::filter(
!(tolower(alias) != tolower(drug_name) &
tolower(alias) %in% tolower(drugAliasPrimary$alias))) |>
## consider unambiguous drug aliases only
dplyr::group_by(alias) |>
dplyr::summarise(
alias_source = paste(unique(alias_source), collapse="|"),
drug_id = paste(unique(drug_id), collapse=","),
.groups = "drop") |>
dplyr::filter(!stringr::str_detect(drug_id,",")) |>
dplyr::distinct() |>
dplyr::mutate(
alias = stringi::stri_enc_toascii(alias)
) |>
dplyr::mutate(
alias_lc = tolower(alias)) |>
## remove aliases that co-incide with ordinary
## english words (pubchem)
dplyr::left_join(
words::words,
by = c("alias_lc" = "word"),
multiple = "all", relationship = "many-to-many") |>
dplyr::filter(
is.na(word_length) |
(!is.na(word_length) &
word_length > 7 &
(alias_source != "pubchem" ))) |>
dplyr::anti_join(
custom_name_ignore, by = "alias_lc") |>
dplyr::select(-c(word_length,alias_lc)) |>
dplyr::distinct() |>
dplyr::select(drug_id, alias, alias_source) |>
dplyr::mutate(
drug_id = as.integer(drug_id)
)
)
return(drugAliasAll)
}
get_mesh_drug_categories <- function(path_data_raw = NULL){
mesh_drug_data <- whatamesh::read_mesh_file(
file.path(path_data_raw,"data-raw",
"mesh","d2022.bin")
)
i <- 1
all_drugs <- data.frame()
while(i <= nrow(mesh_drug_data)){
drugname <- mesh_drug_data[i,]$MH
if(!is.null(mesh_drug_data[i,]$PA)){
drug_category <- paste(unlist(mesh_drug_data[i,]$PA),
collapse=";")
df <- data.frame(
'drugname' = drugname,
'category' = drug_category,
stringsAsFactors = F
)
all_drugs <- all_drugs |>
dplyr::bind_rows(df)
}
i <- i + 1
}
}
sigven/oncoPharmaDB documentation built on Dec. 24, 2024, 9:46 p.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
## Read URL
#'
#' Function that retrieves PubChem compound properties using PubChem public user gateway (PUG)
#'
#' @param q_url Query URL
#' @return a data frame with the following columns of chemical compound properties:
#'
#'
readUrl <- function(q_url) {
out <- tryCatch(
{
readLines(con=q_url, warn=FALSE)
},
error=function(cond) {
# Choose a return value in case of error
return(NA)
},
warning=function(cond) {
# Choose a return value in case of warning
return(NULL)
},
finally={
#message(paste("Processed URL:", q_url))
#message("Some other message at the end")
}
)
return(out)
}
### NCI DRUG DISPLAY LABELS
process_nci_labels <- function(path_data_raw, overwrite = F) {
nci_thesaurus_labels_filepath = file.path(path_data_raw, "nci_thesaurus", "ThesaurusLabels.tsv")
nci_thesaurus_owl_filepath <- file.path(path_data_raw, "nci_thesaurus", "ThesaurusInferred.owl")
if(!is.null(nci_thesaurus_owl_filepath) &
file.exists(nci_thesaurus_owl_filepath) &
(!file.exists(nci_thesaurus_labels_filepath) | overwrite == T)){
system(paste0('/Users/sigven/miniconda3/bin/python data-raw/parse_owl_labels.py ',nci_thesaurus_owl_filepath,' > ',file.path(path_data_raw,'nci_thesaurus','ThesaurusLabels.tsv')))
}
nci_labels <- read.table(file = nci_thesaurus_labels_filepath, sep="\t",comment.char="",
stringsAsFactors = F,quote="",header=F) |>
magrittr::set_colnames(c('nci_t','nci_cd_name','cui')) |>
dplyr::filter(!is.na(nci_t))
return(nci_labels)
}
### CHEMBL-PUBCHEM COMPOUND CROSS-REFERENCE
get_chembl_pubchem_xref <- function(datestamp = '20241024',
chembl_release = "v34",
path_data_raw = NULL,
update = F){
chembl_pubchem_xref_fname <- file.path(
path_data_raw, "chembl",
paste0("chembl_pubchem_mapping_", datestamp,".txt.gz"))
if(!file.exists(chembl_pubchem_xref_fname) | update == T){
download.file(
"ftp://ftp.ebi.ac.uk/pub/databases/chembl/UniChem/data/wholeSourceMapping/src_id1/src1src22.txt.gz",
destfile = chembl_pubchem_xref_fname)
}
chembl_pubchem_compound_xref <- read.table(
gzfile(chembl_pubchem_xref_fname), header = F, skip = 1,
stringsAsFactors = F,sep="\t",quote="") |>
magrittr::set_colnames(c('molecule_chembl_id','pubchem_cid')) |>
dplyr::mutate(chembl_db_version = chembl_release)
return(chembl_pubchem_compound_xref)
}
### TARGETED ANTICANCER COMPOUNDS FROM OPEN TARGETS
get_otp_cancer_drugs <-
function(path_data_raw = NULL,
ot_version = "2024.09"){
cancer_terms <- list()
cancer_terms[['all']] <- phenOncoX::get_terms(
cache_dir = path_data_raw
)
cancer_terms[['efo']] <- cancer_terms[['all']]$records |>
dplyr::select(cui, cui_name, efo_id, primary_site) |>
dplyr::filter(!is.na(efo_id)) |>
dplyr::distinct() |>
dplyr::rename(disease_efo_id = efo_id)
fname <- paste0(path_data_raw,
paste0("/opentargets/opentargets_drugs_",
ot_version,".rds"))
ot_compounds <- as.data.frame(
readRDS(file = fname) |>
dplyr::select(target_genename,
target_symbol,
target_type,
target_ensembl_gene_id,
target_entrezgene,
disease_efo_id,
disease_efo_label,
drug_name,
drug_type,
drug_max_ct_phase,
drug_max_phase_indication,
molecule_chembl_id,
drug_action_type,
drug_moa,
drug_tradenames,
drug_synonyms,
drug_clinical_source,
drug_clinical_id,
drug_description,
drug_withdrawn,
drug_approved_indication,
drug_blackbox_warning,
drug_year_first_approval)) |>
dplyr::mutate(
disease_efo_id = stringr::str_replace_all(
disease_efo_id, "_", ":")) |>
dplyr::distinct() |>
dplyr::left_join(
cancer_terms[['efo']],
by = "disease_efo_id",
multiple = "all",
relationship = "many-to-many") |>
## general cancer ontology terms (neoplasm, cancer, carcinoma, squamous cell carcinoma)
dplyr::mutate(
drug_cancer_relevance =
dplyr::if_else(
!is.na(primary_site) |
disease_efo_id == 'EFO:0000616' |
disease_efo_id == 'EFO:0000311' |
disease_efo_id == 'EFO:0000313' |
disease_efo_id == 'MONDO:0004992' |
disease_efo_id == 'MONDO:0002120' |
disease_efo_id == 'EFO:0003826' |
disease_efo_id == 'EFO:0000707' |
disease_efo_id == 'EFO:0000228',
"by_cancer_condition_otp", "by_other_condition_otp")
) |>
dplyr::mutate(
drug_cancer_relevance = dplyr::if_else(
is.na(disease_efo_id) &
(drug_type == "Antibody" |
drug_type == "Unknown" |
drug_type == "Small molecule") &
!is.na(target_genename) &
stringr::str_detect(
tolower(target_genename),
paste0(
"(kinase|epidermal growth factor|tubulin|",
"oncogene|fibroblast growth factor|",
"dna topoisomerase|tnf receptor superfamily|",
"protein tyrosine phosphatase|",
"receptor tyrosine kinase|",
"platelet derived growth factor)"
)) &
(drug_action_type == "INHIBITOR" |
drug_action_type == "BINDING AGENT" |
drug_action_type == "ANTAGONIST"),
"by_cancer_target_otp",
as.character(drug_cancer_relevance)
)
) |>
dplyr::mutate(
drug_type = dplyr::if_else(
drug_type == "unknown",
"Unknown",
as.character(drug_type)
)
) |>
# dplyr::filter(
# !is.na(drug_action_type)
# ) |>
# dplyr::filter(
# !stringr::str_detect(
# tolower(drug_action_type),
# "allosteric|modulator|substrate|releasing|opener"
# )
# ) |>
dplyr::filter(
drug_type != "Cell" &
drug_type != "Enzyme" &
drug_type != "Gene" &
drug_type != "Oligonucleotide" &
drug_type != "Oligosaccharide"
)
cancer_indication_stats <- ot_compounds |>
dplyr::filter(
!is.na(drug_cancer_relevance) &
drug_cancer_relevance == "by_cancer_condition_otp") |>
dplyr::select(drug_name, disease_efo_label) |>
dplyr::filter(!is.na(disease_efo_label)) |>
dplyr::distinct() |>
dplyr::group_by(drug_name) |>
dplyr::summarise(n_cancer_indications = dplyr::n())
non_cancer_indication_stats <- ot_compounds |>
dplyr::filter(
!is.na(drug_cancer_relevance) &
drug_cancer_relevance == "by_other_condition_otp") |>
#dplyr::filter(drug_cancer_relevance == F) |>
dplyr::filter(!is.na(disease_efo_label)) |>
dplyr::select(drug_name, disease_efo_label,
drug_approved_indication) |>
dplyr::distinct() |>
dplyr::group_by(drug_name) |>
dplyr::summarise(n_noncancer_indications = dplyr::n(),
approved_non_cancer = paste(
unique(drug_approved_indication),
collapse = ";"
), .groups = "drop") |>
dplyr::mutate(approved_non_cancer = dplyr::if_else(
stringr::str_detect(approved_non_cancer,"TRUE"),
as.logical(TRUE),
as.logical(FALSE)
))
indication_stats <- cancer_indication_stats |>
dplyr::full_join(non_cancer_indication_stats, by = "drug_name") |>
dplyr::mutate(n_cancer_indications = dplyr::if_else(
is.na(n_cancer_indications),
as.numeric(0),
as.numeric(n_cancer_indications)
)) |>
dplyr::mutate(n_noncancer_indications = dplyr::if_else(
is.na(n_noncancer_indications),
as.numeric(0),
as.numeric(n_noncancer_indications)
)) |>
dplyr::mutate(drug_n_indications = n_cancer_indications +
n_noncancer_indications) |>
dplyr::mutate(drug_frac_cancer_indications = dplyr::if_else(
n_cancer_indications == 0,
as.numeric(0),
round(as.numeric(n_cancer_indications /
(n_cancer_indications + n_noncancer_indications)),
digits = 3)
)) |>
dplyr::mutate(drug_approved_noncancer = dplyr::if_else(
is.na(approved_non_cancer),
FALSE,
as.logical(approved_non_cancer)
)) |>
dplyr::select(drug_name,
drug_n_indications,
drug_approved_noncancer,
drug_frac_cancer_indications)
ot_compounds <- ot_compounds |>
dplyr::left_join(indication_stats, by = "drug_name") |>
dplyr::mutate(
drug_n_indications = dplyr::if_else(
is.na(drug_n_indications) |
drug_n_indications == "NA",
0,
as.integer(drug_n_indications)
)
)
targeted_compounds <- as.data.frame(
ot_compounds |>
dplyr::distinct() |>
dplyr::group_by_at(
dplyr::vars(-c(drug_clinical_id))) |>
dplyr::summarise(
drug_clinical_id = paste(
unique(drug_clinical_id),
collapse = ","),
drug_clinical_source = paste(
unique(sort(drug_clinical_source)),
collapse = ","),
.groups = "drop") |>
dplyr::ungroup() |>
dplyr::mutate(
drug_name_lc = tolower(drug_name))
)
## adjust max ct phase
drugs_with_max_phase_adj <- as.data.frame(
targeted_compounds |>
#dplyr::filter(drug_cancer_relevance == T) |>
dplyr::filter(!is.na(drug_max_ct_phase) &
!is.na(drug_max_phase_indication)) |>
dplyr::group_by(drug_name, molecule_chembl_id) |>
dplyr::summarise(drug_max_ct_phase = max(drug_max_phase_indication),
.groups = "drop")
)
targeted_compounds <- as.data.frame(targeted_compounds |>
dplyr::select(-drug_max_ct_phase) |>
dplyr::left_join(
drugs_with_max_phase_adj,
by = c("drug_name","molecule_chembl_id"),
multiple = "all", relationship = "many-to-many")
)
targeted_cancer_compounds <- targeted_compounds |>
dplyr::filter(
!is.na(drug_cancer_relevance) &
(drug_cancer_relevance == "by_cancer_condition_otp" |
drug_cancer_relevance == "by_cancer_target_otp")) |>
dplyr::mutate(drug_n_indications = dplyr::if_else(
is.na(drug_n_indications),as.numeric(0),
as.numeric(drug_n_indications)
)) |>
dplyr::mutate(drug_frac_cancer_indications = dplyr::if_else(
is.na(drug_frac_cancer_indications),as.numeric(0),
as.numeric(drug_frac_cancer_indications)
)) |>
dplyr::mutate(drug_approved_noncancer = dplyr::if_else(
is.na(drug_approved_noncancer),as.logical(FALSE),
as.logical(drug_approved_noncancer)
)) |>
dplyr::filter(drug_withdrawn == F)
targeted_noncancer_compounds <- targeted_compounds |>
# dplyr::filter(
# !is.na(drug_cancer_relevance) &
# drug_cancer_relevance == "by_other_condition_otp") |>
dplyr::anti_join(
dplyr::select(
targeted_cancer_compounds, molecule_chembl_id
), by = "molecule_chembl_id"
) |>
dplyr::filter(drug_withdrawn == F)
targeted_noncancer_compounds$drug_name_lc <- NULL
targeted_cancer_compounds$drug_name_lc <- NULL
ot_recs <- dplyr::bind_rows(
targeted_cancer_compounds,
targeted_noncancer_compounds) |>
dplyr::distinct()
return(ot_recs)
}
get_atc_drug_classification <- function(
path_data_raw = NULL){
atc_codes <- as.data.frame(
readr::read_csv(
file = file.path(
path_data_raw, "atc",
"ATC.csv.gz"),
show_col_types = F)) |>
janitor::clean_names() |>
dplyr::mutate(class_id = stringr::str_replace(
class_id, "http://purl.bioontology.org/ontology/ATC/",""
)) |>
dplyr::mutate(parents = stringr::str_replace(
parents, "http://purl.bioontology.org/ontology/ATC/",""
)) |>
dplyr::mutate(
is_drug_class = dplyr::if_else(
is.na(is_drug_class),
FALSE,
TRUE
)
) |>
dplyr::select(class_id, preferred_label, synonyms, obsolete,
parents, atc_level, is_drug_class) |>
dplyr::arrange(atc_level, class_id, parents)
#dplyr::filter(stringr::str_detect(class_id,"^L"))
atc_level1 <- atc_codes |>
dplyr::filter(atc_level == 2) |>
dplyr::rename(atc_level1 = preferred_label,
atc_code_level1 = class_id) |>
dplyr::select(atc_code_level1, atc_level1)
atc_level2 <- atc_codes |>
dplyr::filter(atc_level == 3) |>
dplyr::rename(atc_level2 = preferred_label,
atc_code_level2 = class_id, parent_code = parents) |>
dplyr::select(atc_code_level2, atc_level2, parent_code)
atc_level3 <- atc_codes |>
dplyr::filter(atc_level == 4) |>
dplyr::rename(atc_level3 = preferred_label,
atc_code_level3 = class_id,
parent_code = parents) |>
dplyr::select(atc_code_level3,
atc_level3, parent_code)
atc_level4 <- atc_codes |>
dplyr::filter(atc_level == 5) |>
dplyr::rename(atc_level4 = preferred_label,
atc_code_level4 = class_id,
parent_code = parents) |>
dplyr::select(atc_code_level4,
atc_level4,
parent_code)
atc_drug_classification <- atc_level1 |>
dplyr::left_join(
atc_level2, by = c("atc_code_level1" = "parent_code")) |>
dplyr::left_join(
atc_level3, by = c("atc_code_level2" = "parent_code")
) |>
dplyr::left_join(
atc_level4, by = c("atc_code_level3" = "parent_code")
) |>
dplyr::rename(atc_drug_entry = atc_level4) |>
dplyr::mutate(atc_level2 = stringr::str_replace(
atc_level2,
"in ATC|, ANTINEOPLASTIC DRUGS|, ANTINEOPLASTIC AND IMMUNOMODULATING AGENTS",""
)) |>
dplyr::mutate(atc_level3 = stringr::str_replace(
atc_level3,
"in ATC",""
)) |>
dplyr::mutate(atc_level2 = stringr::str_replace(
atc_level2, "AND ANTIBODY DRUG CONJUGATES","AND ADCs"
)) |>
dplyr::mutate(atc_level3 = dplyr::case_when(
atc_code_level3 == "L03AX" ~ "Other immunostimulants",
atc_code_level3 == "L01CD" ~ "Taxanes",
atc_code_level3 == "L04AX" ~ "Other immunosuppressants",
atc_code_level3 == "L04AB" ~ "TNF-alpha inhibitors",
atc_code_level3 == "L02BX" ~ "Other hormone antagonists",
atc_code_level3 == "L02AE" ~ "GnRH analogs",
atc_code_level3 == "L01XK" ~ "PARP inhibitors",
atc_code_level3 == "L01XX" ~ "Other antineoplastic agents",
atc_code_level3 == "L02AX" ~ "Other hormones for endocrine therapy",
atc_code_level3 == "L02BB" ~ "Anti-androgens",
atc_code_level3 == "L01XH" ~ "HDAC inhibitors",
atc_code_level3 == "L01XG" ~ "Proteasome inhibitors",
atc_code_level3 == "L01XA" ~ "Platinum compounds",
atc_code_level3 == "L01FX" ~ "Other MaBs and ADCs",
atc_code_level3 == "L01FF" ~ "PD-1/PDL-1 inhibitors",
atc_code_level3 == "L01FG" ~ "VEGFR inhibitors",
atc_code_level3 == "L01FE" ~ "EGFR inhibitors",
atc_code_level3 == "L01FD" ~ "HER2 inhibitors",
atc_code_level3 == "L01FC" ~ "CD38 inhibitors",
atc_code_level3 == "L01FB" ~ "CD22 inhibitors",
atc_code_level3 == "L01FA" ~ "CD20 inhibitors",
atc_code_level3 == "L01EX" ~ "Other protein kinase inhibitors",
atc_code_level3 == "L01EN" ~ "FGFR inhibitors",
atc_code_level3 == "L01EM" ~ "PI3K inhibitors",
atc_code_level3 == "L01EL" ~ "BTK inhibitors",
atc_code_level3 == "L01EK" ~ "VEGFR inhibitors",
atc_code_level3 == "L01EJ" ~ "JAK inhibitors",
atc_code_level3 == "L01EH" ~ "HER2 inhibitors",
atc_code_level3 == "L01EG" ~ "mTOR inhibitors",
atc_code_level3 == "L01EF" ~ "CDK inhibitors",
atc_code_level3 == "L01EE" ~ "MEK inhibitors",
atc_code_level3 == "L01ED" ~ "ALK inhibitors",
atc_code_level3 == "L01EC" ~ "BRAF inhibitors",
atc_code_level3 == "L01EB" ~ "EGFR inhibitors",
atc_code_level3 == "L01EA" ~ "BCR-ABL inhibitors",
atc_code_level3 == "L01DC" ~ "Other cytotoxic antibiotics",
atc_code_level3 == "L01CX" ~ "Other plant alkaloids and natural products",
atc_code_level3 == "L01CE" ~ "TOP1 inhibitors",
atc_code_level3 == "L01CC" ~ "Colchicine derivatives",
atc_code_level3 == "L01CB" ~ "Podophyllotoxin derivatives",
atc_code_level3 == "L01CA" ~ "Vinca alkaloids and analogues",
atc_code_level3 == "L01BC" ~ "Pyrimidine analogues",
atc_code_level3 == "L01BB" ~ "Purine analogues",
atc_code_level3 == "L01BA" ~ "Folic acid analogues",
atc_code_level3 == "L01AX" ~ "Other alkylating agents",
atc_code_level3 == "L01XB" ~ "Methylhydrazines",
atc_code_level3 == "L01AG" ~ "Epoxides",
atc_code_level3 == "L01AD" ~ "Nitrosoureas",
atc_code_level3 == "L01AC" ~ "Ethylene imines",
TRUE ~ as.character(atc_level3)
)) |>
dplyr::select(-atc_code_level4) |>
dplyr::distinct()
#dplyr::mutate(atc_drug_entry = dplyr::if_else(
#atc_code_level3 == "L01EX" |
#atc_code_level3 == "L01FX" |
#atc_code_level3 == "L01XX",
#as.character(NA),
#as.character(atc_drug_entry)
#))
atc_custom <- as.data.frame(
readr::read_csv(
file.path(
path_data_raw,
"atc",
"custom_classification.csv"
), col_names = T, show_col_types = F, na = c(".")
)
)
atc_drug_classification <- atc_drug_classification |>
dplyr::bind_rows(atc_custom) |>
dplyr::arrange(atc_code_level3) |>
dplyr::distinct() |>
dplyr::mutate(atc_code_level3 = dplyr::case_when(
## JAK inhibitors (assigned to both L01EJ and L04AF,
## stick to L01EJ for consistency with other kinase inhibitors)
atc_code_level3 == "L04AF" ~ "L01EJ",
## mTOR inhibitors (assigned to both L04AH and L01EG,
## stick to L01EG for consistency with other kinase inhibitors)
atc_code_level3 == "L04AH" ~ "L01EG",
atc_drug_entry == "ipilimumab" |
atc_drug_entry == "tremelimumab" ~ "L01FXA",
atc_drug_entry == "epcoritamab" |
atc_drug_entry == "glofitamab" |
atc_drug_entry == "mosunetuzumab" |
atc_drug_entry == "talquetamab" |
atc_drug_entry == "teclistamab" ~ "L01FXE",
TRUE ~ as.character(atc_code_level3)
)) |>
dplyr::mutate(atc_level3 = dplyr::case_when(
atc_code_level3 == "L01EJ" ~ "JAK inhibitors",
atc_code_level3 == "L01EG" ~ "mTOR inhibitors",
atc_drug_entry == "ipilimumab" |
atc_drug_entry == "tremelimumab" ~ "Other ICIs - CTLA4 inhibitors",
atc_drug_entry == "epcoritamab" |
atc_drug_entry == "glofitamab" |
atc_drug_entry == "mosunetuzumab" |
atc_drug_entry == "talquetamab" |
atc_drug_entry == "teclistamab" ~ "T-cell Engagers",
TRUE ~ as.character(atc_level3)
))
return(atc_drug_classification)
}
## NCI THESAURUS CANCER DRUGS/TREATMENTS
get_nci_drugs <- function(nci_db_release = nci_db_release,
overwrite = F,
path_data_raw = NULL,
path_data_processed = NULL){
nci_antineo_thesaurus <- NULL
nci_drugs <- NULL
nci_ftp_base <- paste0(
"https://evs.nci.nih.gov/ftp1/NCI_Thesaurus/archive/",
nci_db_release,
"_Release/")
nci_thesaurus_files <- list()
nci_thesaurus_files[['flat']] <-
paste0("Thesaurus_", nci_db_release,".FLAT.zip")
nci_thesaurus_files[['owl']] <-
paste0("Thesaurus_", nci_db_release,".OWL.zip")
nci_thesaurus_files[['inf_owl']] <-
paste0("ThesaurusInf_", nci_db_release,".OWL.zip")
options(timeout = 50000)
for (elem in c('flat','owl','inf_owl')) {
remote_file <- paste0(
nci_ftp_base, nci_thesaurus_files[[elem]])
local_file <- file.path(
path_data_raw,"nci_thesaurus",
nci_thesaurus_files[[elem]])
if (!file.exists(local_file)) {
download.file(
url = remote_file,
destfile = local_file,
quiet = T)
system(paste0('unzip -d ',
file.path(path_data_raw, "nci_thesaurus"),
' -o -u ',local_file))
}
}
antineo_agents_url <-
'https://evs.nci.nih.gov/ftp1/NCI_Thesaurus/Drug_or_Substance/Antineoplastic_Agent.txt'
antineo_agents_local <-
file.path(
path_data_raw,
"nci_thesaurus",
"Antineoplastic_Agent.txt")
if (!file.exists(antineo_agents_local)) {
download.file(
url = antineo_agents_url,
destfile = antineo_agents_local, quiet = T)
}
if(!file.exists(
file.path(
path_data_processed,
"nci_thesaurus",
"nci_treatment_thesaurus_antineo_v2.rds")) | overwrite == T){
sorafenib_definition <-
'A synthetic compound targeting growth signaling and angiogenesis. Sorafenib blocks the enzyme RAF kinase, a critical component of the RAF/MEK/ERK signaling pathway that controls cell division and proliferation; in addition, sorafenib inhibits the VEGFR-2/PDGFR-beta signaling cascade, thereby blocking tumor angiogenesis.'
nci_display_labels <- process_nci_labels(
path_data_raw = path_data_raw,
overwrite = overwrite)
## Agents/compounds marked as antineplastic according to NCI
nci_antineo_agents <-
read.table(file = file.path(path_data_raw,"nci_thesaurus","Antineoplastic_Agent.txt"),
header = T, stringsAsFactors = F, comment.char = "",
quote = "", sep = "\t") |>
janitor::clean_names() |>
dplyr::rename(nci_t = code) |>
dplyr::select(nci_t) |>
dplyr::mutate(antineoplastic_agent = TRUE) |>
dplyr::distinct()
## parse all entries in nci thesaurus where the
## semantic concept type is treatment-related
nci_antineo_thesaurus_raw <- as.data.frame(
read.table(file = file.path(path_data_raw, "nci_thesaurus", "Thesaurus.txt"),
header = F, stringsAsFactors = F, sep="\t",
comment.char="", quote = "") |>
dplyr::rename(nci_t = V1, nci_concept_name = V2,
nci_t_parent = V3, concept_synonym = V4,
nci_concept_definition = V5,
nci_cd_name = V6, nci_concept_status = V7,
nci_concept_semantic_type = V8) |>
dplyr::filter(
stringr::str_detect(
nci_concept_semantic_type,
"Chemical|Substance|Therapeutic|Drug|Immunologic")) |>
dplyr::filter(
nci_concept_semantic_type != "Drug Delivery Device" &
nci_concept_semantic_type != "Chemical Viewed Functionally"
) |>
dplyr::filter(
!stringr::str_detect(
tolower(nci_concept_definition), "surgical"
)
) |>
dplyr::filter(
nci_concept_semantic_type != "Therapeutic or Preventive Procedure" |
(nci_concept_semantic_type == "Therapeutic or Preventive Procedure" &
stringr::str_detect(
tolower(concept_synonym), "/"
) &
!stringr::str_detect(
tolower(nci_concept_definition), "regimen"
)
)
) |>
dplyr::left_join(
nci_antineo_agents, by = c("nci_t"),
multiple = "all", relationship = "many-to-many") |>
dplyr::select(-c(nci_concept_name, nci_cd_name)) |>
dplyr::mutate(nci_concept_synonym_all = concept_synonym) |>
tidyr::separate_rows(concept_synonym,sep="\\|") |>
dplyr::mutate(nci_concept_synonym = tolower(concept_synonym)) |>
dplyr::select(-concept_synonym) |>
dplyr::filter(
nci_t != 'C147908' & nci_t != 'C71622') |> ## Hormone Therapy Agent
dplyr::mutate(
nci_concept_semantic_type =
dplyr::if_else(
nci_t == 'C61948','Pharmacologic Substance',
nci_concept_semantic_type)) |> ## redundant Sorafenib entries
dplyr::mutate(
nci_concept_definition =
dplyr::if_else(
nci_t == 'C61948', sorafenib_definition,
nci_concept_definition)) |> ## redundant Sorafenib entries
dplyr::distinct() |>
dplyr::left_join(
nci_display_labels, by = c("nci_t"),
multiple = "all", relationship = "many-to-many") |>
dplyr::filter(
!(nci_t == "C1806" & nci_concept_synonym == "gemtuzumab")) |>
dplyr::filter(
!(nci_t == "C405" & nci_concept_synonym == "ctx")) |>
dplyr::mutate(nci_db_version = nci_db_release) |>
dplyr::filter(!is.na(nci_cd_name)) |>
dplyr::filter(!stringr::str_detect(
tolower(nci_concept_definition), "coronavirus")) |>
dplyr::filter(!stringr::str_detect(
nci_concept_synonym_all,
"SARS-CoV-2|COVID-19|CoV-19|Coronary|Corona|Covid-19|covid-19")) |>
#))
dplyr::mutate(antineoplastic_agent = dplyr::if_else(
is.na(antineoplastic_agent),
as.logical(TRUE),
as.logical(antineoplastic_agent)
)) |>
dplyr::filter(!stringr::str_detect(nci_cd_name," (Gel|Oil|Cream|Seed|Block|Field|Supplement|Factor)$")) |>
dplyr::filter(!stringr::str_detect(nci_cd_name,"(Vaccination|Lotion|Therapeutic Heat|Procedure|Rehabilitation|Prevention|Rinse)$")) |>
dplyr::filter(!stringr::str_detect(nci_cd_name,"(Epitope|Exract|Influenza|Ginseng|Ointment|Management|Injection|Tool)$")) |>
dplyr::filter(!stringr::str_detect(nci_cd_name,"^(Vitamin A Compound|Inactivated Poliovirus|Antineoplastic Immune Cell|Topical)")) |>
dplyr::filter(!stringr::str_detect(nci_cd_name,"^(Sheng-Yu|Ginseng|Dry Cleaning|Boost|Tobacco|Microwave)")) |>
dplyr::filter(!stringr::str_detect(nci_cd_name,"^(Blood|Interruption of|Gum Arabic|Vaginal Cylinder|Laser Ablation|Wheatgrass)")) |>
dplyr::filter(!stringr::str_detect(nci_cd_name,"^(Chemical Challenge|Prevention of|Magic Mouthwash|Wood Dust|Soot|Cocaine)")) |>
dplyr::filter(!stringr::str_detect(nci_cd_name,"^(Antibody|Antigen|Antioxidant|Vaccination|Acetate|Antiserum|Asbestos|Aspirate|Autoantigen|Cytokine)$")) |>
dplyr::filter(!stringr::str_detect(nci_cd_name," Spray| Extract| Antidiabetic| Implant|(Green Tea|Living Healthy|Pollutant|Probe|Protective Agent|Supportive Care|Caffe)"))
#dplyr::filter(
# !stringr::str_detect(
# tolower(nci_concept_definition),
# "chinese |antidiabet|diabetes|antidepress|analgesic|pulmonary edema|nutritional|human carcinogen|anesthetic|nonsedating|sedative|antihyper|antiinflamma|antiarrythm|antiangin|antihist|muscle|neurotransmitter"))
)
nci_antineo_thesaurus <- nci_antineo_thesaurus_raw
## add apatinib as an alias (is missing in NCI thesaurus)
apatinib_alias_entry <- nci_antineo_thesaurus |>
dplyr::filter(nci_concept_synonym == "rivoceranib") |>
dplyr::mutate(nci_concept_synonym = "apatinib") |>
dplyr::mutate(nci_concept_synonym_all =
paste(nci_concept_synonym_all,"Apatinib",sep="|"))
## add nab-paclitaxel as an alias (is missing in NCI thesaurus)
paclitaxel_entry <- nci_antineo_thesaurus |>
dplyr::filter(nci_cd_name == "Paclitaxel") |>
dplyr::select(nci_t, nci_t_parent, nci_concept_definition,
nci_concept_status, nci_concept_semantic_type,
nci_concept_synonym_all, cui) |>
dplyr::distinct()
nab_paclitaxel_entry <- nci_antineo_thesaurus |>
dplyr::filter(nci_cd_name == "Nab-paclitaxel")
nab_paclitaxel_entry$nci_t <- paclitaxel_entry$nci_t
nab_paclitaxel_entry$nci_t_parent <- paclitaxel_entry$nci_t_parent
nab_paclitaxel_entry$cui <- paclitaxel_entry$cui
nab_paclitaxel_entry$nci_concept_definition <- paclitaxel_entry$nci_concept_definition
nab_paclitaxel_entry$nci_concept_status <- paclitaxel_entry$nci_concept_status
nab_paclitaxel_entry$nci_concept_semantic_type <- paclitaxel_entry$nci_concept_semantic_type
nab_paclitaxel_entry$nci_cd_name <- "Paclitaxel"
aliases_paclitaxel <- stringr::str_split(
paclitaxel_entry$nci_concept_synonym_all, "\\|")[[1]]
aliases_nab_paclitaxel <- stringr::str_split(
nab_paclitaxel_entry$nci_concept_synonym_all[1], "\\|")[[1]]
all_paclitaxel_aliases <-
paste(sort(unique(c(aliases_paclitaxel, aliases_nab_paclitaxel))),
collapse="|")
nab_paclitaxel_entry$nci_concept_synonym_all <-
all_paclitaxel_aliases
paclitaxel_entries <- nci_antineo_thesaurus |>
dplyr::filter(nci_cd_name == "Paclitaxel")
paclitaxel_entries$nci_concept_synonym_all <-
all_paclitaxel_aliases
nci_antineo_thesaurus <- nci_antineo_thesaurus |>
dplyr::filter(nci_cd_name != "Paclitaxel" &
nci_cd_name != "Nab-paclitaxel") |>
dplyr::bind_rows(nab_paclitaxel_entry) |>
dplyr::bind_rows(paclitaxel_entries) |>
dplyr::bind_rows(apatinib_alias_entry) |>
dplyr::arrange(nci_cd_name) |>
dplyr::rename(nci_drug_name = nci_concept_synonym) |>
dplyr::filter(!(nci_cd_name == "Sorafenib Tosylate" &
nci_drug_name == "sorafenib")) |>
dplyr::select(-c(nci_t_parent,cui)) |>
dplyr::distinct()
nci_drugs <- list()
## NCI anticancer drugs (non-targeted) - lacking compound identifier (CHEMBL)
nci_drugs[['no_chembl_id']] <- nci_antineo_thesaurus |>
dplyr::select(nci_t,
nci_concept_definition,
nci_cd_name,
nci_drug_name,
nci_concept_synonym_all) |>
dplyr::filter(!stringr::str_detect(
nci_drug_name,
"(material|medication|medicine|medicinal|medical|^was |^agent |no therapy|-line therapy)"
)) |>
dplyr::filter(!stringr::str_detect(
tolower(nci_cd_name),
"(material|medication|medicine|medicinal|medical|^was |^agent |no therapy|-line therapy)"
)) |>
## remove general drug terms ("XX Agent" etc)
dplyr::mutate(num_words = stringr::str_count(
nci_cd_name, " ") + 1) |>
dplyr::filter(
!(num_words <= 4 &
stringr::str_detect(tolower(nci_cd_name), " agent") &
!stringr::str_detect(nci_cd_name, "[0-9]{1,}")
)) |>
dplyr::select(-num_words) |>
dplyr::distinct()
saveRDS(nci_drugs, file = file.path(
path_data_processed, "nci_thesaurus",
"nci_treatment_thesaurus_antineo_v2.rds"))
}else{
nci_drugs <- readRDS(
file = file.path(
path_data_processed, "nci_thesaurus",
"nci_treatment_thesaurus_antineo_v2.rds"))
}
return(nci_drugs)
}
merge_nci_opentargets <- function(
drug_sets = NULL,
path_data_raw = NULL){
## X-ref Open Targets and NCI by drug name
otp_drugs_all <- drug_sets[['otp']] |>
dplyr::mutate(drug_name_lc = tolower(drug_name)) |>
dplyr::left_join(
dplyr::filter(
drug_sets[['nci']][['no_chembl_id']],
!is.na(nci_drug_name)),
by = c("drug_name_lc" = "nci_drug_name"),
multiple = "all", relationship = "many-to-many") |>
dplyr::select(-c("drug_name_lc")) |>
dplyr::distinct() |>
dplyr::mutate(opentargets = TRUE)
## list all drug aliases in targeted drugs from OTP
all_drug_aliases <- otp_drugs_all |>
tidyr::separate_rows(drug_synonyms, sep = "\\|") |>
dplyr::select(drug_synonyms) |>
dplyr::mutate(nci_drug_name = tolower(drug_synonyms)) |>
dplyr::distinct()
## Identify drugs in NCI that are in Open Targets
found_drugs <- drug_sets[['nci']][['no_chembl_id']] |>
dplyr::semi_join(all_drug_aliases,
by = "nci_drug_name") |>
dplyr::select(nci_t, nci_drug_name) |>
dplyr::distinct()
nci_missing <- drug_sets[['nci']][['no_chembl_id']] |>
dplyr::anti_join(found_drugs,
by = "nci_t") |>
dplyr::select(-nci_drug_name) |>
dplyr::mutate(opentargets = F) |>
dplyr::distinct()
## ignore some drugs
custom_name_ignore <- as.data.frame(readr::read_tsv(
file = "data-raw/drug_names_ignore.tsv",
col_names = F, show_col_types = F))
colnames(custom_name_ignore) <- c("nci_cd_name")
## ignore some molecule identifiers (mislabeled)
custom_chembl_ignore <- as.data.frame(readr::read_tsv(
file = "data-raw/molecule_chembl_ids_ignore.tsv",
col_names = F, show_col_types = F))
colnames(custom_chembl_ignore) <- c("molecule_chembl_id")
all_cancer_drugs <- otp_drugs_all |>
dplyr::bind_rows(nci_missing) |>
dplyr::anti_join(
custom_chembl_ignore,
by = "molecule_chembl_id") |>
dplyr::distinct() |>
dplyr::mutate(
nci_cd_name =
dplyr::if_else(
is.na(nci_cd_name) &
!stringr::str_detect(drug_name,"[0-9]"),
Hmisc::capitalize(tolower(drug_name)),
nci_cd_name)) |>
dplyr::mutate(
nci_cd_name = dplyr::if_else(
is.na(nci_cd_name) &
stringr::str_detect(drug_name,"[0-9]"),
drug_name,
nci_cd_name)) |>
dplyr::mutate(
nci_cd_name = dplyr::if_else(
!is.na(nci_cd_name) &
!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL535",
"Sunitinib",
nci_cd_name)) |>
dplyr::mutate(
nci_cd_name = dplyr::if_else(
!is.na(drug_name) &
!is.na(nci_cd_name) &
!is.na(molecule_chembl_id) &
(molecule_chembl_id == "CHEMBL3545427" |
molecule_chembl_id == "CHEMBL5315068" |
molecule_chembl_id == "CHEMBL2109325" |
molecule_chembl_id == "CHEMBL5316170" |
molecule_chembl_id == "CHEMBL31965" |
molecule_chembl_id == "CHEMBL5314999" |
molecule_chembl_id == "CHEMBL2109402" |
molecule_chembl_id == "CHEMBL1790041" |
molecule_chembl_id == "CHEMBL4650278" |
molecule_chembl_id == "CHEMBL3545401"),
stringr::str_to_title(drug_name),
nci_cd_name)) |>
dplyr::mutate(drug_name = dplyr::if_else(
is.na(drug_name),
nci_cd_name,
drug_name)) |>
dplyr::anti_join(
custom_name_ignore,
by = "nci_cd_name")
rownames(all_cancer_drugs) <- NULL
salt_patterns <-
readr::read_tsv(
file.path(path_data_raw, "salts.tsv"),
show_col_types = F, col_names = F)
salt_patterns_regex <- paste0(
"( (",
paste(salt_patterns$X1, collapse="|"),
"))$")
salt_forms <- all_cancer_drugs |>
dplyr::filter(
stringr::str_detect(
tolower(nci_cd_name),
salt_patterns_regex)) |>
dplyr::filter(!is.na(opentargets_version)) |>
dplyr::mutate(tradename = stringr::str_replace(
nci_cd_name,
salt_patterns_regex,
"")) |>
dplyr::select(tradename, nci_cd_name) |>
dplyr::distinct() |>
dplyr::mutate(is_salt = T) |>
dplyr::inner_join(
dplyr::select(all_cancer_drugs, nci_cd_name),
by = c("tradename" = "nci_cd_name"),
multiple = "all", relationship = "many-to-many") |>
dplyr::distinct() |>
dplyr::select(-tradename)
all_cancer_drugs <- all_cancer_drugs |>
dplyr::left_join(
salt_forms,
by = "nci_cd_name",
multiple = "all",
relationship = "many-to-many") |>
dplyr::mutate(is_salt = dplyr::if_else(
is.na(is_salt),
as.logical(FALSE),
as.logical(is_salt)
)) |>
dplyr::distinct()
## antibody drug conjugates
adc_candidates <- all_cancer_drugs |>
dplyr::filter(
(!is.na(nci_cd_name) &
stringr::str_detect(tolower(nci_cd_name), "mab ")) |
stringr::str_detect(
nci_concept_definition, "ADC|antibody(-| )drug conjugate")) |>
dplyr::mutate(is_adc = TRUE) |>
dplyr::mutate(is_adc = dplyr::if_else(
stringr::str_detect(nci_concept_definition, "ADCC") &
!(stringr::str_detect(
tolower(nci_concept_definition),
"(antibody(-| )drug conjugate)") &
(is.na(drug_type) |
drug_type == "Small molecule" |
drug_type == "Antibody")),
as.logical(FALSE),
as.logical(is_adc)
)) |>
dplyr::mutate(is_adc = dplyr::if_else(
is_adc == F &
!is.na(nci_cd_name) &
stringr::str_detect(
tolower(nci_cd_name), "mab ") &
stringr::str_detect(
tolower(nci_cd_name), "(tin|ine|tan|can|tox)$"
),
TRUE,
as.logical(is_adc)
)) |>
dplyr::filter(!is.na(nci_cd_name)) |>
dplyr::select(nci_cd_name, is_adc) |>
dplyr::distinct()
all_cancer_drugs <- all_cancer_drugs |>
dplyr::left_join(
adc_candidates,
by = "nci_cd_name",
multiple = "all",
relationship = "many-to-many") |>
dplyr::mutate(is_adc = dplyr::if_else(
is.na(is_adc),
as.logical(FALSE),
as.logical(is_adc)
)) |>
dplyr::filter(drug_withdrawn == F | is.na(drug_withdrawn)) |>
dplyr::mutate(nci_concept_definition = dplyr::if_else(
!is.na(nci_concept_definition) &
nchar(nci_concept_definition) == 0,
as.character(NA),
as.character(nci_concept_definition)
)) |>
dplyr::mutate(drug_n_indications = dplyr::if_else(
is.na(drug_n_indications),as.numeric(0),
as.numeric(drug_n_indications)
)) |>
dplyr::mutate(drug_frac_cancer_indications = dplyr::if_else(
is.na(drug_frac_cancer_indications),as.numeric(0),
as.numeric(drug_frac_cancer_indications)
)) |>
dplyr::mutate(drug_approved_noncancer = dplyr::if_else(
is.na(drug_approved_noncancer),as.logical(FALSE),
as.logical(drug_approved_noncancer)
)) |>
dplyr::mutate(drug_cancer_relevance = dplyr::if_else(
is.na(drug_cancer_relevance) &
opentargets == F,
"by_nci",
as.character(drug_cancer_relevance)
)) |>
dplyr::distinct()
return(all_cancer_drugs)
}
map_curated_targets <- function(gene_info = NULL,
path_data_raw = NULL,
drug_df = NULL){
drug_df$target_entrezgene <- as.numeric(
drug_df$target_entrezgene
)
drug_target_patterns <-
read.table(file = file.path(
path_data_raw,
"custom_drug_target_regex_nci.tsv"),
sep = "\t", header = T,
stringsAsFactors = F, quote = "") |>
dplyr::inner_join(
gene_info, by = "symbol", multiple = "all",
relationship = "many-to-many") |>
dplyr::distinct()
drugname_suffix <-
paste0("(ab|al|an|at|cl|co|da|de|dy|ea|ed|el|en|er|es|ex|fa|ib|ic|",
"id|il|im|in|ir|is|le|ls|lt|me|na|ne|ns|nt|od|ol|on|or|pt|",
"py|ra|rd|re|rm|rt|se|ta|te|ts|ue|um|us|yl)$")
drug_inhibitor_regex <-
paste0(drugname_suffix,"|^(anti-|inhibitor of)|inhibitor")
all_inhibitors_no_target <- drug_df |>
dplyr::filter(is.na(target_symbol)) |>
dplyr::filter(
stringr::str_detect(
tolower(nci_cd_name),
"inhibitor|antagonist|antibody|blocker|sepantronium| mimetic") |
stringr::str_detect(
tolower(nci_cd_name),
drug_inhibitor_regex) |
(stringr::str_detect(nci_concept_definition,"KRAS") &
stringr::str_detect(nci_concept_definition,"inhibitor"))) |>
dplyr::filter(!stringr::str_detect(
nci_cd_name,
" CAR T|(T|t)herapy|SARS-CoV-2|( (R|r)egimen|(A|a)cid|Hydrochloride)$")) |>
dplyr::filter(!stringr::str_detect(
nci_concept_definition,
"SARS-CoV-2")) |>
dplyr::filter(nchar(nci_concept_definition) > 0)
custom_nci_targeted_drugs <- data.frame()
for(i in 1:nrow(drug_target_patterns)){
pattern <- drug_target_patterns[i, "pattern"]
target_symbol <- drug_target_patterns[i, "symbol"]
target_genename <- drug_target_patterns[i, "genename"]
target_entrezgene <- drug_target_patterns[i, "target_entrezgene"]
target_type <- drug_target_patterns[i, "target_type"]
target_ensembl_gene_id <- drug_target_patterns[i, "target_ensembl_gene_id"]
#target_uniprot_id <- drug_target_patterns[i, "target_uniprot_id"]
hits <- all_inhibitors_no_target |>
dplyr::filter(stringr::str_detect(
nci_cd_name,
pattern = pattern) |
(stringr::str_detect(
tolower(nci_cd_name),
drug_inhibitor_regex) &
stringr::str_detect(
nci_concept_definition, pattern))
)
if(nrow(hits) > 0){
for(n in 1:nrow(hits)){
hit <- hits[n,]
if(stringr::str_detect(tolower(hit$nci_cd_name),
"(mab|art)$|monoclonal antibody|^anti-")){
hit$drug_type <- "Antibody"
}else{
hit$drug_type <- "Small molecule"
}
hit$drug_action_type <- "INHIBITOR"
if(stringr::str_detect(
tolower(hit$nci_cd_name),
"antagonist")){
hit$drug_action_type <- "ANTAGONIST"
}
if(stringr::str_detect(
tolower(hit$nci_cd_name),
"blocker")){
hit$drug_action_type <- "BLOCKER"
}
hit$target_symbol <- target_symbol
hit$target_genename <- target_genename
hit$target_type <- target_type
hit$target_entrezgene <- target_entrezgene
hit$target_ensembl_gene_id <- target_ensembl_gene_id
hit$drug_clinical_source <- "nci_thesaurus_custom"
hit$drug_cancer_relevance <- "by_cancer_target_nci"
hit$drug_n_indications <- 0
hit$drug_frac_cancer_indications <- 0
hit$drug_approved_noncancer <- FALSE
## set general indications for unknown cases
if(is.na(hit$disease_efo_id) &
is.na(hit$disease_efo_label) &
is.na(hit$cui) &
is.na(hit$cui_name)){
hit$disease_efo_id = "EFO:0000311"
hit$disease_efo_label = "cancer"
hit$cui = "C0006826"
hit$cui_name = "Malignant neoplastic disease"
}
custom_nci_targeted_drugs <- custom_nci_targeted_drugs |>
dplyr::bind_rows(hit)
}
}
}
### CHECK HOW MANY TARGET-LACKING INHIBITORS ARE MISSING
### FROM THE CUSTOM NCI MATCHING ROUTINE
inhibitors_no_target_nonmapped <- all_inhibitors_no_target |>
dplyr::anti_join(custom_nci_targeted_drugs, by = "nci_cd_name") |>
dplyr::filter(!stringr::str_detect(
nci_concept_definition,
"(A|a)ntibody(-| )drug conjugate \\(ADC\\)"
)) |>
dplyr::filter(!stringr::str_detect(
tolower(nci_cd_name),"^(allogeneic|regimen |copper |fluorine f |indium |iodine |carbon c|autologous |recombinant |lutetium |yttrium |y 90)|vaccine$"
)) |>
dplyr::filter(
stringr::str_detect(
nci_concept_definition, "antineoplastic|tumor|cancer"
)
) |>
dplyr::select(nci_cd_name,
nci_concept_definition) |>
dplyr::distinct()
drug_df$target_entrezgene <-
as.integer(drug_df$target_entrezgene)
custom_nci_targeted_drugs$target_entrezgene <-
as.integer(custom_nci_targeted_drugs$target_entrezgene)
ot_nci_drugs_curated <-
dplyr::anti_join(drug_df, custom_nci_targeted_drugs,
by = "nci_cd_name") |>
dplyr::bind_rows(custom_nci_targeted_drugs) |>
dplyr::arrange(target_symbol, nci_cd_name) |>
dplyr::mutate(drug_action_type = dplyr::if_else(
(stringr::str_detect(tolower(nci_cd_name),"inhibitor") &
is.na(drug_action_type)) |
(!is.na(nci_cd_name) &
stringr::str_detect(nci_cd_name,"(mab|art)$") &
is.na(drug_action_type)),
"INHIBITOR",
as.character(drug_action_type))) |>
dplyr::mutate(drug_cancer_relevance = dplyr::if_else(
is.na(drug_cancer_relevance) &
(stringr::str_detect(
tolower(nci_concept_definition),
"anti-tumor|chemotherapy|cancer vaccine|immunothera|monoclonal antibody|antineoplastic|treatment of cancer|treatment of metastat") |
stringr::str_detect(tolower(nci_cd_name)," regimen|recombinant|carcinoma|immune checkpoint|anti-programmed cell death ")),
as.character("by_cancer_definition_nci"),
as.character(drug_cancer_relevance)
)) |>
dplyr::mutate(drug_action_type = dplyr::if_else(
is.na(drug_action_type) &
stringr::str_detect(
drug_action_type,
"^(SUBSTRATE|HYDROLYTIC ENZYME|RELEASING AGENT)"),
paste0(drug_action_type,"_OTHER"),
as.character(drug_action_type)
))
duplicated_drugs <- ot_nci_drugs_curated |>
dplyr::group_by(nci_cd_name) |>
dplyr::summarise(drug_cancer_relevance = paste(
sort(unique(drug_cancer_relevance)), collapse="@")) |>
dplyr::filter(stringr::str_detect(drug_cancer_relevance,"@")) |>
tidyr::separate_rows(drug_cancer_relevance, sep = "@") |>
dplyr::filter(!stringr::str_detect(drug_cancer_relevance,"otp"))
ot_nci_drugs_curated <- ot_nci_drugs_curated |>
dplyr::anti_join(duplicated_drugs,
by = c("nci_cd_name","drug_cancer_relevance"))
return(list('curated' = ot_nci_drugs_curated,
'nonmapped' = inhibitors_no_target_nonmapped))
}
assign_drug_category <- function(drug_df = NULL,
path_data_raw = NULL){
atc_classification <-
get_atc_drug_classification(path_data_raw = path_data_raw)
atc_classification_with_drugs <- atc_classification |>
dplyr::mutate(atc_drug_entry = dplyr::case_when(
!is.na(atc_drug_entry) &
(tolower(atc_drug_entry) == "sotorasib" |
tolower(atc_drug_entry) == "ivosidenib" |
tolower(atc_drug_entry) == "enasidenib" |
tolower(atc_drug_entry) == "venetoclax" |
tolower(atc_drug_entry) == "adagrasib") ~ as.character(NA),
TRUE ~ as.character(atc_drug_entry)
)) |>
dplyr::filter(!is.na(atc_drug_entry))
atc_classification_clean <- atc_classification |>
dplyr::filter(is.na(atc_drug_entry)) |>
dplyr::bind_rows(atc_classification_with_drugs) |>
dplyr::select(-atc_drug_entry) |>
dplyr::distinct() |>
dplyr::filter(!is.na(atc_code_level3))
classified_drugs <- list()
classified_drugs[['pre_classified_atc']] <- drug_df |>
dplyr::mutate(drug_entry = tolower(nci_cd_name)) |>
dplyr::left_join(
dplyr::select(
atc_classification_with_drugs,
atc_drug_entry,
atc_code_level3),
by = c("drug_entry" = "atc_drug_entry"),
relationship = "many-to-many") |>
dplyr::filter(!is.na(atc_code_level3)) |>
dplyr::distinct()
drugs_non_classified <- drug_df |>
dplyr::mutate(drug_entry = tolower(nci_cd_name)) |>
dplyr::anti_join(
classified_drugs[['pre_classified_atc']],
by = "drug_entry") |>
dplyr::distinct()
custom_target_classifications <- drugs_non_classified |>
dplyr::filter(!is.na(target_symbol)) |>
dplyr::group_by(drug_name, nci_cd_name,
drug_entry,
nci_concept_definition,
molecule_chembl_id, drug_action_type) |>
dplyr::reframe(
target_symbol = paste(
sort(unique(target_symbol)), collapse="|")) |>
dplyr::mutate(atc_code_level3 = dplyr::case_when(
target_symbol == "ABL1" | target_symbol == "ABL1|BCR" ~ "L01EA",
target_symbol == "BRAF" ~ "L01EC",
stringr::str_detect(target_symbol,"TOP1(\\|TOP2)?") ~ "L01CE",
stringr::str_detect(
target_symbol, "^((PSM(A|B|C|D)[0-9]{1,})|ADRM1|SEM1)") ~ "L01XG",
stringr::str_detect(
target_symbol, "^((BRAF\\|(KDR|RAF))|(ARAF\\|BRAF))") ~ "L01EXJ",
target_symbol == "PLK1" | target_symbol == "PLK4" ~ "L01EXK",
target_symbol == "EGFR" &
!is.na(drug_name) &
stringr::str_detect(drug_name, "MAB") ~ "L01FE",
target_symbol == "ERBB2" &
!is.na(drug_name) &
stringr::str_detect(drug_name, "MAB") ~ "L01FD",
target_symbol == "ERBB2" &
(is.na(drug_name) |
(!is.na(drug_name) &
!stringr::str_detect(drug_name, "MAB"))) ~ "L01EH",
!is.na(drug_name) &
drug_name == "SIROLIMUS" ~ "L01EG",
target_symbol == "EGFR" &
(is.na(drug_name) |
(!is.na(drug_name) &
!stringr::str_detect(drug_name, "MAB"))) ~ "L01EB",
(!is.na(drug_name) &
stringr::str_detect(drug_name, "ABIVERTINIB")) ~ "L01EB",
target_symbol == "ALK" ~ "L01ED",
(drug_action_type == "INHIBITOR" |
drug_action_type == "ANTAGONIST") &
target_symbol == "CTLA4" ~ "L01FXA",
(drug_action_type == "INHIBITOR" |
drug_action_type == "ANTAGONIST") &
target_symbol == "LAG3" ~ "L01FXB",
(drug_action_type == "INHIBITOR" |
drug_action_type == "ANTAGONIST") &
target_symbol == "TIGIT" ~ "L01FXC",
(drug_action_type == "INHIBITOR" |
drug_action_type == "ANTAGONIST") &
(target_symbol == "CD274|CTLA4" |
target_symbol == "CTLA4|PDCD1") ~ "L01FXD",
(!is.na(drug_name) & drug_name == "BRENTUXIMAB VEDOTIN") |
(stringr::str_detect(
target_symbol,"^(TUBA|TUBB)") &
(drug_action_type == "INHIBITOR" |
drug_action_type == "DISRUPTING AGENT")) ~ "L01XXK",
stringr::str_detect(
target_symbol, "^(PARP[1-9]{1}(\\|)?){1,}$") ~ "L01XK",
stringr::str_detect(
target_symbol, "^(HDAC[0-9]{1,}(\\|)?){1,}$") ~ "L01XH",
stringr::str_detect(
target_symbol, "^MAP2K[0-9]") ~ "L01EE",
stringr::str_detect(
target_symbol, "^CDK[0-9]{1,}") ~ "L01EF",
stringr::str_detect(
target_symbol, "^(BIRC|XIAP)") ~ "L01XXF",
stringr::str_detect(
target_symbol, "^JAK[1-2]") ~ "L01EJ",
stringr::str_detect(
target_symbol, "^AURK(A|B|C)") ~ "L01EXB",
stringr::str_detect(
target_symbol, "EGFR") &
stringr::str_detect(
target_symbol,"ERBB") ~ "L01EXH",
stringr::str_detect(
target_symbol, "^FGFR[1-4]{1}") ~ "L01EN",
stringr::str_detect(
target_symbol, "^IDH[1-2]{1}") ~ "L01XXD",
stringr::str_detect(
target_symbol, "^((IGF1\\|)|IGF1R)") ~ "L01XXN",
!is.na(drug_name) &
stringr::str_detect(
tolower(drug_name), "^aberaterone") ~ "L02BX",
!is.na(drug_name) &
stringr::str_detect(
tolower(drug_name), "^(gemcitabine)") ~ "L01BC",
!is.na(drug_name) &
stringr::str_detect(
tolower(drug_name), "^(fludarabine)") ~ "L01BB",
stringr::str_detect(
target_symbol, "^(K|N|H)RAS") ~ "L01XXC",
stringr::str_detect(
target_symbol, "^MET$") ~ "L01EXA",
stringr::str_detect(
target_symbol, "^(CD3D\\|CD3E\\|CD3G)") ~ "L01FXE",
stringr::str_detect(
target_symbol, "^(AKT[0-9](\\|)?){1,}") |
(!is.na(drug_name) &
stringr::str_detect(
drug_name, "GSK-690693")) ~ "L01EXC",
stringr::str_detect(
target_symbol, "^ATR$") ~ "L01XXI",
stringr::str_detect(
target_symbol, "^PDGFR(A|B)|^KIT") ~ "L01EXE",
stringr::str_detect(
target_symbol, "^(NTRK[0-9]\\|?){1,}$") |
(!is.na(drug_name) & drug_name == "TALETRECTINIB") ~ "L01EXF",
stringr::str_detect(
target_symbol, "^(CHEK(1|2)\\|?){1,2}$") ~ "L01XXJ",
stringr::str_detect(
target_symbol, "^BCL2") ~ "L01XXM",
stringr::str_detect(
target_symbol, "^GNRH") ~ "L02AE",
stringr::str_detect(
target_symbol, "^BRD(T|[1-9]{1})") ~ "L01XXA",
stringr::str_detect(
target_symbol, "^MTOR\\|PIK3") ~ "L01XXG",
stringr::str_detect(
target_symbol, "^PIK3") |
(!is.na(drug_name) &
(stringr::str_detect(drug_name,"UMBRALISIB") |
(drug_name == "OMIPALISIB" |
drug_name == "PF-04691502"))) ~ "L01EM",
stringr::str_detect(
target_symbol, "^BTK$") ~ "L01EL",
(!is.na(drug_name) &
(drug_name == "PRALSETINIB" |
drug_name == "SELPERCATINIB")) ~ "L01EXL",
stringr::str_detect(
target_symbol, "^MTOR$") ~ "L01EG",
stringr::str_detect(
target_symbol, "^(KDR|FLT1|FLT3|FLT4)") ~ "L01EK",
stringr::str_detect(
target_symbol, "^(MS4A1)") ~ "L01FA",
stringr::str_detect(
target_symbol, "^(ESR1)") ~ "L02BA",
stringr::str_detect(
target_symbol, "^(CD38)") ~ "L01FC",
stringr::str_detect(
target_symbol, "^(CD22)") ~ "L01FB",
target_symbol == "MAPK1|MAPK3" ~ "L01XXE",
stringr::str_detect(target_symbol, "^AR$") &
(!is.na(nci_concept_definition) &
stringr::str_detect(
nci_concept_definition, "androgen receptor")) ~ "L02BB",
(target_symbol == "PDCD1" |
target_symbol == "CD274") ~ "L01FF",
TRUE ~ as.character(NA)
)) |>
dplyr::mutate(atc_code_level3 = dplyr::if_else(
(!is.na(drug_name) &
stringr::str_detect(
drug_name,
paste0(
"DOVITINIB|BRIVANIB|UCN-01|AT-9283|",
"SURUFATINIB|ORANTINIB|LUCITANIB|",
"ALTIRATINIB|CEP-11981|CRENOLANIB|",
"FORETINIB|FAMITINIB|MOTESANIB|",
"LESTAURTINIB|OSI-930|PACRITINIB|KW-2449|",
"CABOZANTINIB|BMS-817378|BMS-794833|",
"GOLVATINIB|TAK-593|XL-820|TANDUTINIB|IMATINIB|",
"CERDULATINIB|CEP-2563|FEDRATINIB|IBCASERTIB|",
"VANDETANIB|TESEVATINIB|TARLOXOTINIB|",
"TAK-285|SKLB1028|PUQUITINIB|KBP5209|REPOTRECTINIB|",
"VATALANIB|QUIZARTINIB|BMS-690514|IMATINIB|DASATINIB|",
"CANERTINIB|CEP-32496|REGORAFENIB|GUSACITINIB|",
"SU-014813|X-82|XL-999|LINIFANIB|NINGETINIB|",
"PEXIDARTINIB|RG-1530|SITRAVATINIB|SORAFENIB|SUNITINIB|",
"ENTRECTINIB"))) |
(!is.na(nci_concept_definition) &
stringr::str_detect(
tolower(nci_concept_definition),
paste0(
"multikinase|multi(-)?targeted|multi-kinase inhibitor|",
"targets multiple|inhibitor of multiple|multiple-receptor")) &
stringr::str_count(target_symbol, pattern = "\\|") >= 2),
"L01EXI",
as.character(atc_code_level3)
)) |>
dplyr::left_join(
dplyr::select(
atc_classification_clean,
atc_code_level3,
),
by = "atc_code_level3",
relationship = "many-to-many"
) |>
#dplyr::filter(!is.na(atc_code_level3)) |>
dplyr::select(
#molecule_chembl_id,
drug_entry,
atc_code_level3
) |>
dplyr::group_by(drug_entry) |>
dplyr::summarise(
atc_code_level3 = paste(
unique(atc_code_level3), collapse="|")) |>
dplyr::distinct() |>
dplyr::mutate(
atc_code_level3 = stringr::str_replace(
atc_code_level3, "^(NA\\|)|(\\|(NA|L01XXK))$",""
)
)
classified_drugs[['classified_targeted_custom']] <-
drugs_non_classified |>
dplyr::filter(!is.na(target_symbol)) |>
dplyr::left_join(
custom_target_classifications, by = "drug_entry",
relationship = "many-to-many") |>
dplyr::distinct()
classified_drugs_all<-
classified_drugs[['pre_classified_atc']] |>
dplyr::bind_rows(
classified_drugs[['classified_targeted_custom']]) |>
dplyr::bind_rows(
dplyr::filter(drugs_non_classified, is.na(target_symbol))) |>
dplyr::mutate(atc_code_level3 = dplyr::case_when(
(is.na(atc_code_level3) | atc_code_level3 == "NA") &
(!is.na(nci_concept_definition) &
stringr::str_detect(
tolower(nci_concept_definition),
"anthracycline|anthracenedione")) ~ "L01DB",
(is.na(atc_code_level3) | atc_code_level3 == "NA") &
stringr::str_detect(tolower(drug_entry),"hydroxyurea|leucovorin|eniluracil") ~ "L01BA",
(is.na(atc_code_level3) | atc_code_level3 == "NA") &
stringr::str_detect(tolower(drug_entry),"arsenic trioxide") ~ "L01BA",
(is.na(atc_code_level3) | atc_code_level3 == "NA") &
stringr::str_detect(tolower(drug_entry),
"thioguanine|aspacytarabine|aspacytarabine|troxacitabine") ~ "L01BB",
(is.na(atc_code_level3) | atc_code_level3 == "NA") &
!stringr::str_detect(drug_entry,"/| ") &
stringr::str_detect(drug_entry, "xel$") ~ "L01CD",
(is.na(atc_code_level3) | atc_code_level3 == "NA") &
!stringr::str_detect(drug_entry,"/| ") &
stringr::str_detect(drug_entry, "platin$") ~ "L01XA",
(is.na(atc_code_level3) | atc_code_level3 == "NA") &
!is.na(nci_concept_definition) &
stringr::str_detect(tolower(nci_concept_definition), "anti-estrogen") ~ "L02BA",
(is.na(atc_code_level3) | atc_code_level3 == "NA") &
!is.na(nci_concept_definition) &
stringr::str_detect(tolower(nci_concept_definition), "aromatase inhibitor") ~ "L02BG",
(is.na(atc_code_level3) | atc_code_level3 == "NA") &
!is.na(nci_concept_definition) &
stringr::str_detect(tolower(nci_concept_definition), "nitrogen mustard") ~ "L01AA",
(is.na(atc_code_level3) | atc_code_level3 == "NA") &
stringr::str_detect(
tolower(nci_concept_definition), "purine( nucleoside)? analog") ~ "L01BB",
(is.na(atc_code_level3) | atc_code_level3 == "NA") &
stringr::str_detect(
tolower(nci_concept_definition), "pyrimidine( nucleoside)? analog") ~ "L01BC",
(is.na(atc_code_level3) | atc_code_level3 == "NA") &
!is.na(drug_entry) &
!stringr::str_detect(drug_entry, "/") &
stringr::str_detect(
tolower(nci_concept_definition), "vinca alkaloid") ~ "L01CA",
(is.na(atc_code_level3) | atc_code_level3 == "NA") &
((!is.na(nci_concept_definition) &
stringr::str_detect(
tolower(nci_concept_definition),
"antineoplastic activit|anti-tumor activit"
)) |
(!is.na(drug_max_ct_phase) &
stringr::str_detect(
drug_entry,"(in|ib|ide|ine|ax|il|an|ate| alfa)$") &
drug_max_ct_phase >= 2 &
(!is.na(drug_n_indications) &
drug_n_indications > 2) &
(!is.na(drug_frac_cancer_indications) &
drug_frac_cancer_indications > 0.4))) ~ "L01XX",
TRUE ~ as.character(atc_code_level3)
)) |>
dplyr::distinct() |>
dplyr::group_by(dplyr::across(-c("atc_code_level3"))) |>
dplyr::summarise(atc_code_level3 = paste(unique(atc_code_level3), collapse="|"),
.groups = "drop") |>
dplyr::mutate(atc_code_level3 = dplyr::case_when(
stringr::str_detect(
atc_code_level3, "^(L0[A-Z0-9]{1,}\\|((S|M)01(XA|LA|AH)))$") ~
stringr::str_replace_all(
atc_code_level3, "(\\|((S|M)01(XA|LA|AH)))$", ""),
stringr::str_detect(
atc_code_level3, "L0[A-Z0-9]{1,}\\|L01XX") ~
stringr::str_replace_all(
atc_code_level3, "\\|L01XX", ""),
atc_code_level3 == "D11AH|L01EJ" |
atc_code_level3 == "D11AH|L01XF" |
atc_code_level3 == "G03DA|L02AB" |
atc_code_level3 == "D10AD|L01XF" |
atc_code_level3 == "D11AH|L04AD" ~
stringr::str_replace_all(
atc_code_level3, "^(D10AD|G03DA|D11AH)\\|", ""),
atc_code_level3 == "L01FG|S01LA" |
atc_code_level3 == "L01EG|S01XA" |
atc_code_level3 == "L01BB|L04AA" |
atc_code_level3 == "L01BA|L04AX" |
atc_code_level3 == "L01XX|M01AH" ~
stringr::str_replace_all(
atc_code_level3, "\\|(S01LA|S01XA|L04AA|L04AX|M01AH)", ""),
TRUE ~ atc_code_level3
)) |>
dplyr::distinct()
atc_classified_drugs <- classified_drugs_all |>
dplyr::filter(!is.na(drug_entry)) |>
dplyr::select(atc_code_level3, drug_entry) |>
dplyr::filter(!is.na(atc_code_level3)) |>
tidyr::separate_rows(atc_code_level3, sep = "\\|") |>
dplyr::filter(!is.na(atc_code_level3)) |>
dplyr::left_join(
dplyr::select(
atc_classification_clean,
atc_code_level1,
atc_level1,
atc_code_level2,
atc_level2,
atc_code_level3,
atc_level3
), relationship ="many-to-many"
) |>
dplyr::group_by(drug_entry) |>
dplyr::summarise(
atc_code_level1 = paste(unique(atc_code_level1), collapse="|"),
atc_level1 = paste(unique(stringr::str_trim(atc_level1)), collapse="|"),
atc_code_level2 = paste(unique(atc_code_level2), collapse="|"),
atc_level2 = paste(unique(stringr::str_trim(atc_level2)), collapse="|"),
atc_code_level3 = paste(unique(atc_code_level3), collapse="|"),
atc_level3 = paste(unique(stringr::str_trim(atc_level3)), collapse="|")) |>
dplyr::mutate(atc_code_level1 = dplyr::if_else(
!is.na(atc_code_level1),
stringr::str_replace(
atc_code_level1,"(^NA\\|)|\\|NA$",""),
as.character(atc_code_level1)
)) |>
dplyr::mutate(atc_level1 = dplyr::if_else(
!is.na(atc_level1),
stringr::str_replace(
atc_level1,"(^NA\\|)|\\|NA$",""),
as.character(atc_level1)
)) |>
dplyr::mutate(atc_code_level2 = dplyr::if_else(
!is.na(atc_code_level2),
stringr::str_replace(
atc_code_level2,"(^NA\\|)|\\|NA$",""),
as.character(atc_code_level2)
)) |>
dplyr::mutate(atc_level2 = dplyr::if_else(
!is.na(atc_level2),
stringr::str_replace(
atc_level2,"(^NA\\|)|\\|NA$",""),
as.character(atc_level2)
)) |>
dplyr::mutate(atc_code_level3 = dplyr::if_else(
!is.na(atc_code_level3),
stringr::str_replace(
atc_code_level3,"(^NA\\|)|\\|NA$",""),
as.character(atc_code_level3)
)) |>
dplyr::mutate(atc_level3 = dplyr::if_else(
!is.na(atc_level3),
stringr::str_replace(
atc_level3,"(^NA\\|)|\\|NA$",""),
as.character(atc_level3)
)) |>
dplyr::mutate(atc_code_level3 = dplyr::if_else(
!is.na(atc_code_level3),
stringr::str_replace(
atc_code_level3,"L01XX\\|L01XXK","L01XXK"),
as.character(atc_code_level3)
)) |>
dplyr::mutate(atc_level3 = dplyr::if_else(
!is.na(atc_level3),
stringr::str_replace(
atc_level3,
"Other antineoplastic agents\\|Tubulin inhibitors",
"Tubulin inhibitors"),
as.character(atc_level3)
)) |>
dplyr::mutate(atc_level3 = dplyr::if_else(
!is.na(atc_level3),
stringr::str_replace(
atc_level3,
"Folic acid analogues|Other immunosuppressants",
"Folic acid analogues"),
as.character(atc_level3)
))
classified_drugs_all_final <- classified_drugs_all |>
dplyr::select(-atc_code_level3) |>
dplyr::left_join(atc_classified_drugs, by = "drug_entry",
relationship = "many-to-many") |>
dplyr::mutate(atc_treatment_category = "unknown") |>
dplyr::mutate(atc_treatment_category = dplyr::case_when(
!is.na(atc_code_level2) &
stringr::str_detect(
atc_code_level2, "(\\|L02(A|B))|(^L02(A|B))") ~
"cancer_hormone_therapy",
!is.na(target_symbol) &
!is.na(atc_code_level2) &
stringr::str_detect(
atc_code_level2, "(\\|L01(E|F))|(^L01(E|F))"
) ~ "cancer_targeted_therapy",
!is.na(target_symbol) &
!is.na(atc_code_level3) &
stringr::str_detect(
atc_code_level3, "(\\|L01XX[A-J])|(^L01XX[A-J])"
) ~ "cancer_targeted_therapy",
!is.na(target_symbol) &
!is.na(atc_code_level3) &
stringr::str_detect(
atc_code_level3, "(\\|L01XXM)|(^L01XXM)"
) ~ "cancer_targeted_therapy",
!is.na(target_symbol) &
!is.na(atc_code_level3) &
stringr::str_detect(
atc_code_level3, "(\\|L04)|(^L04)"
) ~ "cancer_immuno_suppressants",
!is.na(target_symbol) &
stringr::str_detect(drug_cancer_relevance,"^by_cancer") &
(is.na(atc_code_level3) |
atc_code_level3 == "NA" |
atc_code_level3 == "L01XX")
~ "cancer_unclassified",
!is.na(target_symbol) &
!is.na(atc_code_level3) &
stringr::str_detect(
atc_code_level3, "(\\|L01X[A-L])|(^L01X[A-L])"
) ~ "cancer_targeted_therapy",
!is.na(atc_code_level2) &
!stringr::str_detect(
atc_code_level2, "(\\|L01(E|F))|(^L01(E|F))") &
((!is.na(atc_code_level3) &
drug_type != "Antibody" &
stringr::str_detect(atc_code_level3, "L01XXK")) |
(!is.na(atc_code_level3) &
stringr::str_detect(atc_code_level3,"L01X(A|B)")) |
(stringr::str_detect(
atc_code_level2, "(L01(A|B|C|D))"
))) ~ "cancer_chemo_therapy",
(!is.na(atc_code_level3) &
drug_type == "Antibody" &
stringr::str_detect(
atc_code_level3, "L01XXK")
) ~ "cancer_adc_targeted_therapy",
!is.na(target_symbol) &
!is.na(atc_code_level2) &
atc_code_level2 != "NA" &
!stringr::str_detect(
atc_code_level2, "\\|L0|^L0"
) ~ "other_targeted_therapy",
TRUE ~ as.character(atc_treatment_category)
))
cancer_targets_classified <- classified_drugs_all_final |>
dplyr::filter(
(drug_cancer_relevance == "by_cancer_condition_otp" |
drug_cancer_relevance == "by_cancer_target_otp" |
drug_cancer_relevance == "by_cancer_target_nci") &
atc_treatment_category == "cancer_targeted_therapy") |>
dplyr::select(target_symbol) |>
dplyr::filter(!is.na(target_symbol)) |>
dplyr::filter(!stringr::str_detect(target_symbol,"^(RP(S|L)|TUB)")) |>
dplyr::distinct() |>
dplyr::mutate(cancer_target_classified = TRUE)
classified_drugs_all_final <- classified_drugs_all_final |>
dplyr::left_join(cancer_targets_classified) |>
dplyr::mutate(cancer_target_classified = dplyr::if_else(
is.na(target_symbol) | is.na(cancer_target_classified),
FALSE,
as.logical(cancer_target_classified)
))
drug_df <- remove_duplicate_chembl_ids(
drug_df = classified_drugs_all_final)
drug_df$drug_entry <- NULL
return(drug_df)
}
remove_duplicate_chembl_ids <- function(drug_df = NULL){
black_list <-
readr::read_tsv(
file = "data-raw/drug_name_black_list.txt",
col_names = F, show_col_types = F)
black_list$nci_cd_name <- black_list$X1
drug_df <- drug_df |> dplyr::anti_join(
black_list, by = "nci_cd_name"
)
return(drug_df)
}
clean_final_drug_list <- function(drug_df = NULL){
pharmaoncox <- drug_df |>
dplyr::filter(
is.na(nci_cd_name) |
!stringr::str_detect(
nci_cd_name,
"Sustained-release| Bead(s)?| Compound|Vector|Pegylated")
) |>
dplyr::filter(
is.na(nci_concept_definition) |
!stringr::str_detect(
tolower(nci_concept_definition),
"a( (synthetic|diagnostic|targeted))?( radio(immuno)conjugate)")
) |>
dplyr::filter(
is.na(drug_action_type) |
!stringr::str_detect(tolower(drug_action_type),"vaccine")
) |>
dplyr::filter(
is.na(nci_concept_definition) |
!stringr::str_detect(
tolower(nci_concept_definition),
"(^(a|any)|cancer|tumor|dna|autologous|cell-based|synthetic|cell|peptide|valent)( cell)? vaccine")
) |>
dplyr::filter(
is.na(nci_cd_name) |
!stringr::str_detect(
nci_cd_name,
"^(Carbon C|Fluorine F|Gallium Ga|Indium In|Iodine I|Lutetium Lu|Technetium Tc|Yttrium Y)"
)
) |>
dplyr::filter(is.na(nci_cd_name) |
nci_cd_name != "Abivertinib Maleate") |>
dplyr::mutate(nci_cd_name = dplyr::if_else(
is.na(nci_cd_name),
stringr::str_to_title(drug_name),
as.character(nci_cd_name)
)) |>
dplyr::mutate(is_salt = dplyr::if_else(
nci_cd_name == "Avitinib Maleate",
as.logical(NA),
as.logical(is_salt)
)) |>
dplyr::mutate(drug_name = dplyr::if_else(
nci_cd_name == "Doxycycline" &
!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL1433",
"DOXYCYCLINE",
as.character(drug_name)))
drug_action_types <- as.data.frame(
pharmaoncox |>
dplyr::select(nci_cd_name, drug_action_type) |>
dplyr::distinct() |>
dplyr::group_by(nci_cd_name) |>
dplyr::summarise(drug_action_type = paste(
drug_action_type, collapse = "/"
))
)
pharmaoncox$drug_action_type <- NULL
pharmaoncox <- pharmaoncox |>
dplyr::left_join(
drug_action_types,
by = "nci_cd_name",
relationship = "many-to-many") |>
dplyr::select(drug_name,
nci_cd_name,
drug_type,
drug_action_type,
molecule_chembl_id,
drug_moa,
drug_max_phase_indication,
dplyr::everything())
drug_max_ct_phase <- as.data.frame(
pharmaoncox |>
dplyr::select(nci_cd_name, drug_max_ct_phase) |>
dplyr::group_by(nci_cd_name) |>
dplyr::summarise(drug_max_ct_phase = max(drug_max_ct_phase))
)
pharmaoncox$drug_max_ct_phase <- NULL
pharmaoncox <- pharmaoncox |>
dplyr::left_join(drug_max_ct_phase,
by = "nci_cd_name",
relationship = "many-to-many") |>
dplyr::select(-c(drug_moa)) |>
dplyr::mutate(nci_concept_synonym2 = dplyr::if_else(
is.na(nci_concept_synonym_all) & !is.na(drug_synonyms),
as.character(tolower(drug_synonyms)),
as.character(tolower(nci_concept_synonym_all))
)) |>
dplyr::mutate(nci_concept_synonym_all2 = nci_concept_synonym_all) |>
tidyr::separate_rows(nci_concept_synonym2,
sep="\\|") |>
dplyr::rename(nci_concept_synonym = nci_concept_synonym2) |>
dplyr::select(-c(nci_concept_synonym_all2,
drug_synonyms,
drug_tradenames,
drug_description)) |>
dplyr::distinct() |>
dplyr::select(drug_name, nci_cd_name, drug_type,
drug_action_type, drug_cancer_relevance,
molecule_chembl_id,
drug_max_phase_indication, drug_max_ct_phase,
target_genename, target_symbol,
target_type, target_ensembl_gene_id,
target_entrezgene,
disease_efo_id, disease_efo_label,
cui, cui_name, primary_site,
nci_concept_synonym,
nci_concept_synonym_all,
dplyr::everything()) |>
dplyr::mutate(nci_concept_definition =
stringi::stri_enc_toascii(nci_concept_definition)) |>
dplyr::mutate(nci_concept_synonym_all =
stringi::stri_enc_toascii(nci_concept_synonym_all)) |>
dplyr::mutate(nci_concept_synonym =
stringi::stri_enc_toascii(nci_concept_synonym)) |>
dplyr::mutate(drug_name =
stringi::stri_enc_toascii(drug_name)) |>
dplyr::mutate(
nci_cd_name =
stringi::stri_enc_toascii(nci_cd_name)) |>
dplyr::mutate(drug_action_type = stringr::str_replace_all(
drug_action_type, "/NA|NA/",""
)) |>
dplyr::mutate(idx = dplyr::row_number())
## Simplify records with only "cancer" indications,
## mapping them to a unique
## EFO/CUI cross-ref, avoiding similar records
##with "neoplasm", "carcinoma" etc.
pharmaoncox_cancer_no_indication <- pharmaoncox |>
dplyr::filter(is.na(disease_efo_id))
pharmaoncox_non_cancer <- pharmaoncox |>
dplyr::filter(is.na(primary_site) & !is.na(disease_efo_id) &
drug_cancer_relevance == "by_other_condition_otp")
pharmaoncox_cancer_NOS <- as.data.frame(
pharmaoncox |>
dplyr::filter(is.na(primary_site) & !is.na(disease_efo_id) &
drug_cancer_relevance != "by_other_condition_otp") |>
dplyr::mutate(disease_efo_id = "EFO:0000311",
disease_efo_label = "cancer",
cui = "C0006826",
cui_name = "Malignant neoplastic disease") |>
dplyr::group_by(
dplyr::across(-dplyr::ends_with(c("drug_clinical_id")))) |>
dplyr::summarise(
drug_clinical_id = paste(unique(drug_clinical_id), collapse=","),
.groups = "drop"
) |>
dplyr::distinct()
)
pharmaoncox_cancer_specific <- pharmaoncox |>
dplyr::filter(!is.na(primary_site))
pharmaoncox <- pharmaoncox_cancer_no_indication |>
dplyr::bind_rows(pharmaoncox_non_cancer) |>
dplyr::bind_rows(pharmaoncox_cancer_specific) |>
dplyr::bind_rows(pharmaoncox_cancer_NOS) |>
dplyr::arrange(nci_cd_name) |>
dplyr::mutate(nci_cd_name = dplyr::if_else(
stringr::str_detect(nci_cd_name,";-"),
drug_name,
as.character(nci_cd_name)
)) |>
dplyr::filter(!is.na(nci_cd_name)) |>
dplyr::mutate(drug_name_final = dplyr::case_when(
!is.na(drug_name) &
!is.na(nci_cd_name) &
tolower(nci_cd_name) != tolower(drug_name) &
stringr::str_detect(
tolower(nci_cd_name),"(ib|mab|in)$") &
stringr::str_detect(
drug_name, "[0-9]{1,}") ~ nci_cd_name,
!is.na(drug_name) &
!is.na(nci_cd_name) &
tolower(nci_cd_name) != tolower(drug_name) &
!stringr::str_detect(
tolower(nci_cd_name),"(ib|mab|in)$") &
stringr::str_detect(
drug_name, "[0-9]{1,}") ~ drug_name,
!is.na(drug_name) &
!is.na(nci_cd_name) &
tolower(nci_cd_name) == tolower(drug_name) ~
stringr::str_to_title(tolower(nci_cd_name)),
is.na(drug_name) & !is.na(nci_cd_name) ~ nci_cd_name,
TRUE ~ as.character(
stringr::str_to_title(tolower(drug_name))
)
)) |>
dplyr::filter(
!(drug_name_final == "CFI-400945" &
molecule_chembl_id == "CHEMBL3408947") &
!(drug_name_final == "Relebactam" &
molecule_chembl_id == "CHEMBL3112741") &
!(drug_name_final == "SR16234" &
molecule_chembl_id == "CHEMBL3545210")
)
pharmaoncox$drug_name <- NULL
pharmaoncox$drug_name <- pharmaoncox$drug_name_final
pharmaoncox$nci_cd_name <- NULL
pharmaoncox$drug_name_final <- NULL
pharmaoncox$nci_concept_synonym <- NULL
blackbox_warnings <- as.data.frame(pharmaoncox |>
dplyr::select(drug_name, drug_blackbox_warning) |>
dplyr::distinct() |>
dplyr::group_by(drug_name) |>
dplyr::summarise(drug_blackbox_warning = paste(
unique(drug_blackbox_warning), collapse=","))
)
nci_t_map <- as.data.frame(pharmaoncox |>
dplyr::select(drug_name, nci_t, nci_concept_definition) |>
dplyr::distinct() |>
dplyr::group_by(drug_name) |>
dplyr::summarise(
nci_t = paste(
unique(nci_t), collapse=","),
nci_concept_definition = paste(
unique(nci_concept_definition),
collapse = ",")
), .groups = "drop"
)
pharmaoncox$nci_t <- NULL
pharmaoncox$nci_concept_definition <- NULL
pharmaoncox$drug_blackbox_warning <- NULL
pharmaoncox2 <- pharmaoncox |>
dplyr::left_join(
nci_t_map, by = "drug_name",
relationship = "many-to-many") |>
dplyr::left_join(
blackbox_warnings, by = "drug_name",
relationship = "many-to-many") |>
dplyr::distinct() |>
dplyr::mutate(drug_action_type = dplyr::if_else(
drug_action_type == "NA" &
stringr::str_detect(tolower(drug_name),
"ib |ib$") &
stringr::str_detect(tolower(nci_concept_definition),
"inhibitor"),
"INHIBITOR",
as.character(drug_action_type)
)) |>
dplyr::mutate(drug_type = dplyr::if_else(
drug_action_type == "NA" &
stringr::str_detect(tolower(drug_name),
"ib |ib$") &
stringr::str_detect(tolower(nci_concept_definition),
"inhibitor"),
"Small molecule",
as.character(drug_type)
)) |>
dplyr::mutate(inhibition_moa = dplyr::if_else(
!stringr::str_detect(
drug_action_type,
paste0(
"^(POSITIVE|NEGATIVE|AGONIST|HYDROLYTIC|",
"CHELATING|PARTIAL|NA|INVERSE|OPENER|",
"RELEASING|ANTISENSE|STABILISER|MODULATOR|",
"SEQUESTERING|OXIDATIVE|OTHER|ACTIVATOR|RNA|",
"VACCINE|DEGRADER|PROTEOLYTIC|SUBSTRATE)")) &
(is.na(drug_type) |
(drug_type != "Protein" &
drug_type != "Gene" &
drug_type != "Cell")) &
(is.na(target_type) |
target_type != "protein_protein_interaction"),
as.logical(TRUE),
as.logical(FALSE),
as.logical(FALSE)
)) |>
dplyr::mutate(inhibition_moa = dplyr::if_else(
inhibition_moa == F &
drug_action_type == "OTHER" &
(stringr::str_detect(
tolower(drug_name),"mab$") |
drug_type == "Antibody"),
as.logical(TRUE),
as.logical(inhibition_moa)
))
inhibition_moa_df <- as.data.frame(pharmaoncox2 |>
dplyr::select(drug_name, inhibition_moa) |>
dplyr::distinct() |>
dplyr::group_by(drug_name) |>
dplyr::summarise(inhibition_moa = paste(
unique(inhibition_moa), collapse="|"
), .groups = "drop")
) |>
dplyr::mutate(inhibition_moa = dplyr::if_else(
stringr::str_detect(inhibition_moa, "\\|"),
as.logical(TRUE),
as.logical(inhibition_moa)
))
pharmaoncox2$inhibition_moa <- NULL
pharmaoncox2 <- pharmaoncox2 |>
dplyr::left_join(
inhibition_moa_df,
by = "drug_name",
relationship = "many-to-many") |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL1742994") &
drug_name == "Brentuximab vedotin")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL1908394") &
drug_name == "GSK-461364")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL2158685") &
drug_name == "ABC-294640")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL2331680") &
drug_name == "RG-7603")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL253969") &
drug_name == "OSI-632")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL283120") &
drug_name == "AXL-1717")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL52885") &
drug_name == "ENMD-981693")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL4650827") &
drug_name == "SNDX-5613 FREE BASE")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL1645462") &
drug_name == "AC-480")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL4303241") &
drug_name == "BAY-1161909")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL4298098") &
drug_name == "SAR-408701")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL4091801") &
drug_name == "APG115")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL3623290") &
drug_name == "AZD-3759")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL103") &
drug_name == "Mycophenolic Acid")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL3545007") &
drug_name == "RG-7602")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL3545003") &
drug_name == "MSC-2363318A")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL4297310") &
drug_name == "ONC-201")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL4297930") &
drug_name == "CYC-065")) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL3899477") &
(nci_t == "NA" | is.na(nci_t)))) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL3126004") &
(nci_t == "NA" | is.na(nci_t)))) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL376408") &
(nci_t == "NA" | is.na(nci_t)))) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL4297489") &
(nci_t == "NA" | is.na(nci_t)))) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL17157") &
(nci_t == "NA" | is.na(nci_t)))) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL1201231") &
(nci_t == "NA" | is.na(nci_t)))) |>
dplyr::filter(!((!is.na(molecule_chembl_id) &
molecule_chembl_id == "CHEMBL1433") &
(nci_t == "NA" | is.na(nci_t)))) |>
dplyr::filter(!(drug_name == "Risedronate Sodium" |
drug_name == "Reminertant" |
drug_name == 'Ibandronate Sodium'))
drug_maps <- list()
drug_maps[['id2name']] <- pharmaoncox2 |>
dplyr::select(drug_name) |>
dplyr::distinct() |>
dplyr::mutate(
drug_id = dplyr::row_number())
if(nrow(drug_maps[['id2name']]) != length(
unique(pharmaoncox2$drug_name))){
lgr::lgr$fatal("Ambiguous drug names")
return(0)
}
pharmaoncox2 <- pharmaoncox2 |>
dplyr::left_join(
drug_maps[['id2name']],
by = "drug_name",
multiple = "all")
drug_maps[['id2target']] <- pharmaoncox2 |>
dplyr::select(drug_id,
target_symbol,
target_entrezgene,
target_genename,
target_ensembl_gene_id,
target_type) |>
dplyr::distinct()
drug_maps[['id2indication']] <- pharmaoncox2 |>
dplyr::select(drug_id,
drug_max_phase_indication,
drug_approved_indication,
drug_frac_cancer_indications,
drug_approved_noncancer,
drug_n_indications,
drug_clinical_source,
drug_year_first_approval,
drug_clinical_id,
drug_max_ct_phase,
disease_efo_id,
disease_efo_label,
cui,
cui_name,
primary_site) |>
dplyr::distinct()
drug_maps[['id2synonym']] <- pharmaoncox2 |>
dplyr::select(drug_id,
nci_concept_synonym_all) |>
dplyr::distinct()
drug_maps[['id2basic']] <- as.data.frame(
pharmaoncox2 |>
dplyr::select(
drug_id,
molecule_chembl_id,
drug_type,
drug_action_type,
drug_cancer_relevance,
inhibition_moa,
is_salt,
is_adc,
drug_blackbox_warning,
nci_concept_definition,
nci_t,
opentargets,
atc_code_level1,
atc_level1,
atc_code_level2,
atc_level2,
atc_code_level3,
atc_level3,
atc_treatment_category) |>
dplyr::distinct())
return(drug_maps)
}
expand_drug_aliases <- function(drug_index_map = NULL,
path_data_raw = NULL,
chembl_pubchem_datestamp = chembl_pubchem_datestamp){
chembl_pubchem_xref <-
get_chembl_pubchem_xref(
datestamp = chembl_pubchem_datestamp,
path_data_raw = path_data_raw)
drugAliasPrimary <-
drug_index_map[['id2name']] |>
dplyr::select(drug_id, drug_name) |>
dplyr::mutate(alias = drug_name) |>
dplyr::select(-drug_name) |>
dplyr::mutate(alias_source = "primaryName")
drugAliasNCI <- as.data.frame(
drug_index_map[['id2synonym']] |>
dplyr::select(drug_id, nci_concept_synonym_all) |>
tidyr::separate_rows(nci_concept_synonym_all, sep="\\|") |>
dplyr::distinct() |>
dplyr::rename(alias = nci_concept_synonym_all) |>
dplyr::filter(nchar(alias) > 3) |>
dplyr::mutate(alias_source = "nci")
)
## Extend aliases with those found in PubChem
## get drug set that contains PubChem cross-references
unique_chembl_pubchem <- drug_index_map[['id2basic']] |>
dplyr::select(drug_id, molecule_chembl_id) |>
dplyr::filter(!is.na(molecule_chembl_id)) |>
dplyr::distinct() |>
dplyr::left_join(
chembl_pubchem_xref,
by = "molecule_chembl_id",
multiple = "all", relationship = "many-to-many") |>
dplyr::filter(!is.na(pubchem_cid)) |>
dplyr::select(-c(chembl_db_version))
## Retrieve aliases for drugs with PubChem x-refs
lgr::lgr$info("Retrieving additional drug aliases/synonyms for PubChem-mapped compounds")
pubchem_synonym_files <-
sort(list.files(path = file.path(path_data_raw, "pubchem"),
pattern = "CID-Synonym-filtered_",
full.names = T))
drugAliasPubchem <- data.frame()
for(f in pubchem_synonym_files){
synonym_data <- as.data.frame(readr::read_tsv(
f, col_names = c('pubchem_cid','alias'),
col_types = "dc",
progress = F
))
pubchem_alias_df <- synonym_data |>
dplyr::inner_join(
unique_chembl_pubchem,
by = "pubchem_cid",
multiple = "all", relationship = "many-to-many")
if(nrow(pubchem_alias_df) > 0){
pubchem_alias_df <- pubchem_alias_df |>
dplyr::select(-c(pubchem_cid, molecule_chembl_id)) |>
dplyr::mutate(alias_source = "pubchem")
drugAliasPubchem <-
drugAliasPubchem |>
dplyr::bind_rows(pubchem_alias_df)
}
rm(synonym_data)
}
## ignore some drugs
custom_name_ignore <- as.data.frame(readr::read_tsv(
file = "data-raw/drug_names_ignore.tsv",
col_names = F, show_col_types = F))
colnames(custom_name_ignore) <- c("alias_lc")
custom_name_ignore$alias_lc <- tolower(
custom_name_ignore$alias_lc)
drugAliasAll <- as.data.frame(drugAliasPrimary |>
dplyr::bind_rows(drugAliasNCI) |>
dplyr::bind_rows(drugAliasPubchem) |>
dplyr::filter(nchar(alias) > 3 &
nchar(alias) < 250) |>
#tidyr::separate_rows(alias, sep = "\\|") |>
dplyr::distinct() |>
dplyr::left_join(drug_index_map[['id2name']],
by = "drug_id") |>
## avoid drug aliases for a particular drug that are identical
## to the primary drug name of another drug
dplyr::filter(
!(tolower(alias) != tolower(drug_name) &
tolower(alias) %in% tolower(drugAliasPrimary$alias))) |>
## consider unambiguous drug aliases only
dplyr::group_by(alias) |>
dplyr::summarise(
alias_source = paste(unique(alias_source), collapse="|"),
drug_id = paste(unique(drug_id), collapse=","),
.groups = "drop") |>
dplyr::filter(!stringr::str_detect(drug_id,",")) |>
dplyr::distinct() |>
dplyr::mutate(
alias = stringi::stri_enc_toascii(alias)
) |>
dplyr::mutate(
alias_lc = tolower(alias)) |>
## remove aliases that co-incide with ordinary
## english words (pubchem)
dplyr::left_join(
words::words,
by = c("alias_lc" = "word"),
multiple = "all", relationship = "many-to-many") |>
dplyr::filter(
is.na(word_length) |
(!is.na(word_length) &
word_length > 7 &
(alias_source != "pubchem" ))) |>
dplyr::anti_join(
custom_name_ignore, by = "alias_lc") |>
dplyr::select(-c(word_length,alias_lc)) |>
dplyr::distinct() |>
dplyr::select(drug_id, alias, alias_source) |>
dplyr::mutate(
drug_id = as.integer(drug_id)
)
)
return(drugAliasAll)
}
get_mesh_drug_categories <- function(path_data_raw = NULL){
mesh_drug_data <- whatamesh::read_mesh_file(
file.path(path_data_raw,"data-raw",
"mesh","d2022.bin")
)
i <- 1
all_drugs <- data.frame()
while(i <= nrow(mesh_drug_data)){
drugname <- mesh_drug_data[i,]$MH
if(!is.null(mesh_drug_data[i,]$PA)){
drug_category <- paste(unlist(mesh_drug_data[i,]$PA),
collapse=";")
df <- data.frame(
'drugname' = drugname,
'category' = drug_category,
stringsAsFactors = F
)
all_drugs <- all_drugs |>
dplyr::bind_rows(df)
}
i <- i + 1
}
}
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.