update mpa function

R/trees.R

@@ -1,21 +1,30 @@
 #' Get metphlan tree or data v31
 #'
-#' \code{mpa} gets the metaphlan tree or data v31
+#' \code{mpa} gets the metaphlan tree or data v31. In the case of nodes,
+#' the node_label is the same as taxid.
 #'
-#' @param data TRUE for data. FALSE for tree. Default is FALSE.
+#' @param x A character string: tree, tips, or nodes. Default is tree.
 #'
 #' @return A phylo or data.frame (see the `data` param)
 #' @export
 #'
-mpa <- function(data = FALSE) {
-    if (!data) {
+mpa <- function(x = 'tree') {
+    if (x == 'tree') {
         fname <- system.file(
             'extdata', 'mpav31.newick', package = 'taxPPro'
         )
         output <- ape::read.tree(fname)
-    } else if (data) {
+    } else if (x == 'tips') {
+        fname <- system.file(
+            'extdata', 'mpav31_tips.tsv', package = 'taxPPro'
+        )
+        output <- utils::read.table(
+            file = fname, header = TRUE, sep = '\t',
+            row.names = NULL
+        )
+    } else if (x == 'nodes') {
         fname <- system.file(
-            'extdata', 'mpav31.tsv', package = 'taxPPro'
+            'extdata', 'mpav31_nodes.tsv', package = 'taxPPro'
         )
         output <- utils::read.table(
             file = fname, header = TRUE, sep = '\t',

inst/extdata/mpav31_nodes.tsv

@@ -1,4 +1,4 @@
-"taxid"	"kingdom_taxid"	"phylum_taxid"	"class_taxid"	"order_taxid"	"family_taxid"	"genus_taxid"	"species_taxid"	"strain_taxid"
+"node_label"	"kingdom_taxid"	"phylum_taxid"	"class_taxid"	"order_taxid"	"family_taxid"	"genus_taxid"	"species_taxid"	"strain_taxid"
 "2157"	"2157"	NA	NA	NA	NA	NA	NA	NA
 "1801616"	"2157"	"1801616"	NA	NA	NA	NA	NA	NA
 "28890"	"2157"	"28890"	NA	NA	NA	NA	NA	NA

inst/scripts/get_mpa_tree_v31.R

@@ -147,7 +147,8 @@ nodes_new_taxonomy <- map(nodes_taxonomy, ~ {
         order_taxid, family_taxid, genus_taxid, species_taxid
     ) |>
     discard(~ all(is.na(.x)))
-
+nodes_new_taxonomy <- nodes_new_taxonomy |>
+    rename(node_label = taxid)
 
 # sp_taxid_dups <- mpa_data$species_taxid[which(duplicated(mpa_data$species_taxid))]
 #