Merge pull request #446 from stitam/stralerts

Fix chembl_query() for the compound structural alerts resource

Author: stitam

NEWS.md

@@ -11,6 +11,11 @@
 ## MINOR IMPROVEMENTS
 
 * `chembl_query()` now returns a named list and uses better formatting for nested output.
+* Added new argument `tidy = TRUE` to `chembl_query()` so we can now control whether we want to try to convert the output to a flat format.
+
+## BUG FIXES
+
+* `chembl_query()` did not work with the "compound_structural_alerts" resource. This has been fixed.
 
 # webchem 1.3.1
 

R/chembl.R

@@ -191,6 +191,7 @@ chembl_files <- function(version = "latest") {
 #' @param query character; a vector of ChEMBL IDs.
 #' @param resource character; the ChEMBL resource to query. Use
 #' [chembl_resources()] to see all available resources.
+#' @param tidy logical; attempt to convert output to a simpler structure.
 #' @param cache_file character; the name of the cache file without the file
 #' extension. If \code{NULL}, results are not cached.
 #' @param verbose logical; should a verbose output be printed on the console?
@@ -245,8 +246,8 @@ chembl_files <- function(version = "latest") {
 #' chembl_query("CHEMBL1", resource = "chembl_id_lookup")
 #' # Resource: compound_record - requires record ID
 #' chembl_query("1", resource = "compound_record")
-#' # Resource: compound_structural_alert - requires compound structural alert ID
-#' chembl_query("79048021", resource = "compound_structural_alert")
+#' # Resource: compound_structural_alert - requires compound ChEMBL ID
+#' chembl_query("CHEMBL266429", resource = "compound_structural_alert")
 #' # Resource: document - requires document ChEMBL ID
 #' chembl_query("CHEMBL1158643", resource = "document")
 #' # Resource: document_similarity - requires document 1 ChEMBL ID
@@ -294,6 +295,7 @@ chembl_files <- function(version = "latest") {
 #' @export
 chembl_query <- function(query,
                          resource = "molecule",
+                         tidy = TRUE,
                          cache_file = NULL,
                          similarity = 70,
                          verbose = getOption("verbose"),
@@ -317,17 +319,16 @@ chembl_query <- function(query,
     query <- chembl_validate_query(query, resource, verbose)
     if (is.na(query)) return(NA)
     if (verbose) webchem_message("query", query, appendLF = FALSE)
-    if (resource == "similarity") {
-    url <- ifelse(
-      test_service_down, "",
-      paste0(stem, "/", resource, "/", query, "/", similarity, ".json")
-      )
+    if (test_service_down) {
+      url <- ""
+    } else if (resource == "similarity") {
+      url <- paste0(stem, "/", resource, "/", query, "/", similarity, ".json")
+    } else if (resource == "compound_structural_alert") {
+      url <- paste0(stem, "/", resource, ".json?molecule_chembl_id=", query)
     } else {
-      url <- ifelse(
-      test_service_down, "",
-      paste0(stem, "/", resource, "/", query, ".json")
-      )
+      url <- paste0(stem, "/", resource, "/", query, ".json")
     }
+
     webchem_sleep(type = "API")
     res <- try(httr::RETRY("GET",
                            url,
@@ -344,7 +345,9 @@ chembl_query <- function(query,
     }
     if (verbose) message(httr::message_for_status(res))
     cont <- httr::content(res, type = "application/json")
-    cont <- format_chembl(cont)
+    if (tidy) {
+      cont <- format_chembl(cont)
+    }
     return(cont)
   }
   if (is.null(cache_file)) {
@@ -488,7 +491,6 @@ chembl_validate_query <- function(query, resource, verbose) {
     "activity",
     "binding_site",
     "compund_record",
-    "compound_structural_alert",
     "drug_indication",
     "drug_warning",
     "mechanism",
@@ -506,6 +508,7 @@ chembl_validate_query <- function(query, resource, verbose) {
     "assay",
     "biotherapeutic",
     "chembl_id_lookup",
+    "compound_structural_alert",
     "document",
     "document_similarity",
     "drug",
@@ -621,6 +624,7 @@ format_chembl <- function(cont) {
     "biotherapeutic",
     "biocomponents",
     "chembl_release",
+    "compound_structural_alerts",
     "cross_references",
     "go_slims",
     "indication_refs",
@@ -844,7 +848,7 @@ chembl_example_query <- function(resource) {
     cell_line = c("CHEMBL3307241", "CHEMBL3307242"),
     chembl_id_lookup = "CHEMBL1",
     compound_record = "1",
-    compound_structural_alert = "79048021",
+    compound_structural_alert = "CHEMBL266429",
     document = c("CHEMBL1158643", "CHEMBL1132398", "CHEMBL5303573", "CHEMBL3639173"),
     document_similarity = "CHEMBL1148466",
     drug = "CHEMBL2",

man/chembl_query.Rd

@@ -63,8 +63,8 @@ test_that("chembl_query() examples", {
   o7 <- chembl_query("CHEMBL1", resource = "chembl_id_lookup")
   # Resource: compound_record - requires record ID
   o8 <- chembl_query("1", resource = "compound_record")
-  # Resource: compound_structural_alert - requires compound structural alert ID
-  o9 <- chembl_query("79048021", resource = "compound_structural_alert")
+  # Resource: compound_structural_alert - requires compound ChEMBL ID
+  o9 <- chembl_query("CHEMBL266429", resource = "compound_structural_alert", tidy = FALSE)
   # Resource: document - requires document ChEMBL ID
   o10 <- chembl_query("CHEMBL1158643", resource = "document")
   # Resource: document_similarity - requires document 1 ChEMBL ID
@@ -91,7 +91,8 @@ test_that("chembl_query() examples", {
   # Resource: protein_classification - requires protein class ID
   o22 <- chembl_query("1", resource = "protein_classification")
   # Resource: similarity - requires SMILES
-  o23 <- chembl_query("CC(=O)Oc1ccccc1C(=O)O", resource = "similarity")
+  o23 <- suppressWarnings(
+    chembl_query("CC(=O)Oc1ccccc1C(=O)O", resource = "similarity"))
   # Resource: source - requires source ID
   o24 <- chembl_query("1", resource = "source")
   # Resource: substructure - requires SMILES
@@ -119,7 +120,7 @@ test_that("chembl_query() examples", {
   expect_true(inherits(o6, "list") & length(o6[[1]]) == 11)
   expect_true(inherits(o7, "list") & length(o7[[1]]) == 5)
   expect_true(inherits(o8, "list") & length(o8[[1]]) == 6)
-  #expect_true(inherits(o9, "list") & length(o9[[1]]) == 3)
+  expect_true(inherits(o9, "list") & length(o9[[1]][[1]]) == 7)
   expect_true(inherits(o10, "list") & length(o10[[1]]) == 19)
   expect_true(inherits(o11, "list") & length(o11[[1]]) == 4)
   expect_true(inherits(o12, "list") & length(o12[[1]]) == 30)