Feature/ add ability to fetch latest version of BioGRID (#203)

* updat `get_biogrid_pin() so that it can determine the latest version and download/process it from BioGRID

* update NEWS; bump dev version

DESCRIPTION

@@ -1,7 +1,7 @@
 Package: pathfindR
 Type: Package
 Title: Enrichment Analysis Utilizing Active Subnetworks
-Version: 2.3.1.9002
+Version: 2.3.1.9003
 Authors@R: c(person("Ege", "Ulgen",
                     role = c("cre", "cph"), 
                     email = "[email protected]",

NEWS.md

@@ -8,6 +8,10 @@
 - updated the `get_kegg_gsets()` function to also use `ggkegg` for fetching genes per pathway data
 - removed unneeded dependencies: `magick`, `KEGGgraph` and `KEGGREST`
 
+## Minor Changes and Bug Fixes
+
+- updated the `get_biogrid_pin()` function so that it can now determine the latest version and download/process it from BioGRID (via setting `release = "latest"`, which is now the default behavior) 
+
 # pathfindR 2.3.1
 
 ## Minor Changes and Bug Fixes

R/data_generation.R

@@ -25,12 +25,12 @@ process_pin <- function(pin_df) {
 #' list of available organisms (default = 'Homo_sapiens')
 #' @param path2pin the path of the file to save the PIN data. By default, the
 #' PIN data is saved in a temporary file
-#' @param release the requested BioGRID release (default = '4.4.224')
+#' @param release the requested BioGRID release (default = 'latest')
 #'
 #' @return the path of the file in which the PIN data was saved. If
 #' \code{path2pin} was not supplied by the user, the PIN data is saved in a
 #' temporary file
-get_biogrid_pin <- function(org = "Homo_sapiens", path2pin, release = "4.4.224") {
+get_biogrid_pin <- function(org = "Homo_sapiens", path2pin, release = "latest") {
     # check organism name
     all_org_names <- c("Anopheles_gambiae_PEST", "Apis_mellifera", "Arabidopsis_thaliana_Columbia",
         "Bacillus_subtilis_168", "Bos_taurus", "Caenorhabditis_elegans", "Candida_albicans_SC5314",
@@ -55,6 +55,14 @@ get_biogrid_pin <- function(org = "Homo_sapiens", path2pin, release = "4.4.224")
         stop(paste(org, "is not a valid Biogrid organism.", "Available organisms are listed on: https://wiki.thebiogrid.org/doku.php/statistics"))
     }
 
+    if (release == "latest") {
+      result <- httr::GET("https://downloads.thebiogrid.org/BioGRID/Latest-Release/")
+      result <- httr::content(result, "text")
+
+      h2_matches <- regexpr("(?<=<h2>BioGRID Release\\s)(\\d\\.\\d\\.\\d+)", result, perl = TRUE)
+      release <- regmatches(result, h2_matches)
+    }
+
     # release directory for download
     rel_dir <- paste0("BIOGRID-", release)
 

tests/testthat/test-data_generation.R

@@ -28,13 +28,31 @@ test_that("`get_biogrid_pin()` -- returns a path to a valid PIN file", {
     expected_biogrid_pin_df <- data.frame(V1 = expected_biogrid_pin_df$Interactor_A,
         V2 = "pp", V3 = expected_biogrid_pin_df$Interactor_B)
 
-    pin_path <- get_biogrid_pin()
+    pin_path <- get_biogrid_pin(release = "4.4.211")
     pin_df <- read.delim(pin_path, header = FALSE)
     expect_true(ncol(pin_df) == 3)
     expect_true(all(pin_df[, 2] == "pp"))
     expect_identical(pin_df, expected_biogrid_pin_df)
 })
 
+test_that("`get_biogrid_pin()` -- determines and downloads the latest version", {
+  mockery::stub(get_biogrid_pin, "utils::download.file", NULL)
+  mockery::stub(get_biogrid_pin, "utils::unzip", list(Name = "BIOGRID-ORGANISM-Homo_sapiens-X.X.X.tab3.txt"))
+  mockery::stub(get_biogrid_pin, "utils::read.delim", toy_biogrid_pin)
+
+  expected_biogrid_pin_df <- toy_biogrid_pin
+  colnames(expected_biogrid_pin_df) <- c("Interactor_A", "Interactor_B")
+  expected_biogrid_pin_df <- process_pin(expected_biogrid_pin_df)
+  expected_biogrid_pin_df <- data.frame(V1 = expected_biogrid_pin_df$Interactor_A,
+                                        V2 = "pp", V3 = expected_biogrid_pin_df$Interactor_B)
+
+  pin_path <- get_biogrid_pin()
+  pin_df <- read.delim(pin_path, header = FALSE)
+  expect_true(ncol(pin_df) == 3)
+  expect_true(all(pin_df[, 2] == "pp"))
+  expect_identical(pin_df, expected_biogrid_pin_df)
+})
+
 test_that("`get_biogrid_pin()` -- error check works", {
     # invalid organism error
     expect_error(get_biogrid_pin(org = "Hsapiens"), paste("Hsapiens is not a valid Biogrid organism.",