fixed DO mapping

sigven · May 19, 2023 · 889b403 · 889b403
1 parent bbbce75
commit 889b403
Show file tree

Hide file tree

Showing 3 changed files with 49 additions and 30 deletions.
diff --git a/R/sysdata.rda b/R/sysdata.rda
diff --git a/data-raw/data-raw.R b/data-raw/data-raw.R
@@ -21,7 +21,7 @@ lgr::lgr$appenders$console$set_layout(
 )
 ## Get UMLS / DiseaseOntology / EFO mappings
 umls_map <- map_umls(
-  update = T,
+  update = F,
   basedir = here::here())
 
 icd10_map <- map_icd10(
@@ -40,7 +40,7 @@ efo_map <- map_efo(
   umls_map = umls_map,
   efo_release = metadata$phen_oncox[
     metadata$phen_oncox$source == "Experimental Factor Ontology",]$version,
-  update = T,
+  update = F,
   basedir = here::here())
 
 ## Use OncoTree as starting point for phenotype cross-map
@@ -73,7 +73,7 @@ db[['oncotree_core']] <- oncotree_core
 db[['oncotree_expanded']] <- oncotree_expanded
 db[['auxiliary_maps']] <- auxiliary_maps
 
-version_bumped <- "0.5.9"
+version_bumped <- "0.6.0"
 gd_records <- list()
 db_id_ref <- data.frame()
 

diff --git a/data-raw/phen_oncox_utilities.R b/data-raw/phen_oncox_utilities.R
@@ -799,33 +799,6 @@ map_disease_ontology <- function(
     i <- i + 1
   }
 
-  do_map_cui_name_matched <- do_map |>
-    dplyr::select(do_id, do_name) |>
-    dplyr::left_join(
-      umls_map_lower, 
-      by = c("do_name" = "cui_name_lc"),
-      multiple = "all", relationship = "many-to-many") |>
-    dplyr::filter(!is.na(cui)) |>
-    dplyr::select(do_id, cui) |>
-    dplyr::distinct()
-
-  #do_map <- dplyr::left_join(do_map, do_map_cui_name_matched)
-
-  if (skip_non_cui_mapped == T) {
-    do_map1 <- do_map |> 
-      dplyr::filter(!is.na(cui)) |> 
-      dplyr::distinct()
-    do_map2 <- do_map |> 
-      dplyr::filter(is.na(cui)) |>
-      dplyr::select(-c(cui)) |>
-      dplyr::inner_join(
-        do_map_cui_name_matched, 
-        by = c("do_id"), 
-        multiple = "all", relationship = "many-to-many")
-
-    do_map <- dplyr::bind_rows(do_map1, do_map2)
-  }
-
   ## manual correction of erroneous or missing UMLS cross-references
   do_map <- do_map |>
     # dplyr::mutate(do_name =
@@ -887,6 +860,24 @@ map_disease_ontology <- function(
     ## HER2-receptor negative breast cancer
     dplyr::mutate(cui = dplyr::if_else(do_id == "DOID:0060080",
                                        "C5238910", as.character(cui))) |>
+    ## Diffuse Midline Glioma, H3 K27M-mutant
+    dplyr::mutate(cui = dplyr::if_else(do_id == "DOID:0080684",
+                                       "C4289688", as.character(cui))) |>
+    ## Diffuse large B-cell lymphoma activated B-cell type
+    dplyr::mutate(cui = dplyr::if_else(do_id == "DOID:0080996",
+                                       "C1333296", as.character(cui))) |>
+    ## Oligodendroglioma
+    dplyr::mutate(cui = dplyr::if_else(do_id == "DOID:3181",
+                                       "C0028945", as.character(cui))) |>
+    ## B-cell Acute Lymphoblastic Leukemia
+    dplyr::mutate(cui = dplyr::if_else(do_id == "DOID:0080638",
+                                       "C1292769", as.character(cui))) |>
+    ## B-cell Adult Acute Lymphocytic Leukemia
+    dplyr::mutate(cui = dplyr::if_else(do_id == "DOID:0060592",
+                                       "C0279593", as.character(cui))) |>
+    ## Diffuse Glioma, H3 G34 Mutant
+    dplyr::mutate(cui = dplyr::if_else(do_id == "DOID:0080880",
+                                       "C5669880", as.character(cui))) |>
 
     dplyr::bind_rows(
       data.frame(do_id = "DOID:1788",
@@ -988,6 +979,34 @@ map_disease_ontology <- function(
                  cui = "C0007129", do_cancer_slim = T,
                  stringsAsFactors = F))
 
+
+  do_map_cui_name_matched <- do_map |>
+    dplyr::select(do_id, do_name) |>
+    dplyr::left_join(
+      umls_map_lower, 
+      by = c("do_name" = "cui_name_lc"),
+      multiple = "all", relationship = "many-to-many") |>
+    dplyr::filter(!is.na(cui)) |>
+    dplyr::select(do_id, cui) |>
+    dplyr::distinct()
+
+  #do_map <- dplyr::left_join(do_map, do_map_cui_name_matched)
+
+  if (skip_non_cui_mapped == T) {
+    do_map1 <- do_map |> 
+      dplyr::filter(!is.na(cui)) |> 
+      dplyr::distinct()
+    do_map2 <- do_map |> 
+      dplyr::filter(is.na(cui)) |>
+      dplyr::select(-c(cui)) |>
+      dplyr::inner_join(
+        do_map_cui_name_matched, 
+        by = c("do_id"), 
+        multiple = "all", relationship = "many-to-many")
+
+    do_map <- dplyr::bind_rows(do_map1, do_map2)
+  }
+
   return(do_map)
 }