From d5d8c19799ff38e689a70079c89cad5dcdfe5390 Mon Sep 17 00:00:00 2001
From: Tamas Stirling <stirling.tamas@gmail.com>
Date: Fri, 28 Jun 2024 20:49:52 +0200
Subject: [PATCH] Find BioSample ID regardless of GCA or GCF

---
 DESCRIPTION                     |  2 +-
 NAMESPACE                       |  1 +
 R/ncbi_link.R                   | 43 +++++++++++++++------------------
 man/ncbi_link.Rd                |  2 --
 tests/testthat/test-ncbi_link.R | 43 +++++++++++++++++++++++++++++++++
 5 files changed, 64 insertions(+), 27 deletions(-)
 create mode 100644 tests/testthat/test-ncbi_link.R

diff --git a/DESCRIPTION b/DESCRIPTION
index 95f513a..c963f08 100644
--- a/DESCRIPTION
+++ b/DESCRIPTION
@@ -30,6 +30,6 @@ Suggests:
     knitr,
     rmarkdown,
     testthat
-RoxygenNote: 7.3.1
+RoxygenNote: 7.3.2
 VignetteBuilder: knitr
 Config/testthat/edition: 3
diff --git a/NAMESPACE b/NAMESPACE
index af932ba..2ea93ba 100644
--- a/NAMESPACE
+++ b/NAMESPACE
@@ -13,6 +13,7 @@ export(ncbi_download_genome)
 export(ncbi_get_meta)
 export(ncbi_get_summary)
 export(ncbi_get_uid)
+export(ncbi_link)
 export(ncbi_link_uid)
 export(ncbi_parse)
 export(ncbi_parse_assembly_xml)
diff --git a/R/ncbi_link.R b/R/ncbi_link.R
index 010f1cd..56c482e 100644
--- a/R/ncbi_link.R
+++ b/R/ncbi_link.R
@@ -20,8 +20,7 @@
 #' ncbi_link("GCF_000002435.2", from = "assembly", to = "biosample")
 #' ncbi_link("SAMN02714232", from = "biosample", to = "assembly")
 #' }
-#' 
-#' 
+#' @export
 ncbi_link <- function(
     query, 
     from,
@@ -61,34 +60,30 @@ ncbi_link_assembly_biosample <- function(
     batch_size = batch_size,
     verbose = verbose
   )
-  res <- list()
-  for (i in 1:nrow(from_uid$web_history)) {
-    WH <- list(
-      "WebEnv" = from_uid$web_history$WebEnv[i],
-      "QueryKey" = from_uid$web_history$QueryKey[i]
-    )
-    class(WH) <- c("web_history", "list")
-    hit <- wrap(
-      "entrez_summary",
-      package = "rentrez",
-      verbose = verbose,
-      db = "assembly",
-      web_history = WH
-    )
-    if ("esummary" %in% class(hit)) {
-      hit <- list(hit)
-    }
-    res[[i]] <- hit
-  }
-  res <- unlist(res, recursive = FALSE)
+  res <- ncbi_get_summary(query = from_uid, verbose = verbose)
   ids <- data.frame(
-    assembly = unname(sapply(res, function(x) x$assemblyaccession)),
+    assembly_gbk = unname(sapply(res, function(x) x$synonym$genbank)),
+    assembly_rsq = unname(sapply(res, function(x) x$synonym$refseq)),
     biosample = unname(sapply(res, function(x) x$biosampleaccn))
   )
+  ids$assembly_rsq <- ifelse(ids$assembly_rsq == "", NA, ids$assembly_rsq)
+  index_gbk <- which(ids$assembly_gbk %in% assembly[which(!is.na(assembly))])
+  index_rsq <- which(ids$assembly_rsq %in% assembly[which(!is.na(assembly))])
   out <- data.frame(
     assembly = assembly
   )
-  out <- dplyr::left_join(out, ids, by = "assembly")
+  out_gbk <- dplyr::right_join(
+    out, 
+    ids[index_gbk, c("assembly_gbk", "biosample")], 
+    by = c("assembly" = "assembly_gbk")
+  )
+  out_rsq <- dplyr::right_join(
+    out,
+    ids[index_rsq, c("assembly_rsq", "biosample")],
+    by = c("assembly" = "assembly_rsq")
+  )
+  out_both <- dplyr::bind_rows(out_gbk, out_rsq)
+  out <- dplyr::left_join(out, out_both, by = "assembly")
   out <- tibble::as_tibble(out)
   return(out)
 }
diff --git a/man/ncbi_link.Rd b/man/ncbi_link.Rd
index 3ce60fa..5cc30e4 100644
--- a/man/ncbi_link.Rd
+++ b/man/ncbi_link.Rd
@@ -36,6 +36,4 @@ attempts to link ID-s from one database to another.
 ncbi_link("GCF_000002435.2", from = "assembly", to = "biosample")
 ncbi_link("SAMN02714232", from = "biosample", to = "assembly")
 }
-
-
 }
diff --git a/tests/testthat/test-ncbi_link.R b/tests/testthat/test-ncbi_link.R
new file mode 100644
index 0000000..1ece05b
--- /dev/null
+++ b/tests/testthat/test-ncbi_link.R
@@ -0,0 +1,43 @@
+data(examples)
+
+test_that("ncbi_link() assembly to biosample from GCA",{
+  res <- ncbi_link("GCA_001698945.1", from = "assembly", to = "biosample")
+  
+  expect_equal(class(res), c("tbl_df", "tbl", "data.frame"))
+  expect_equal(dim(res), c(1,2))
+  expect_equal(res$assembly, "GCA_001698945.1")
+  expect_equal(res$biosample, "SAMN03175161")
+})
+
+test_that("ncbi_link() assembly to biosample from GCF",{
+  res <- ncbi_link("GCF_001698945.1", from = "assembly", to = "biosample")
+  
+  expect_equal(class(res), c("tbl_df", "tbl", "data.frame"))
+  expect_equal(dim(res), c(1,2))
+  expect_equal(res$assembly, "GCF_001698945.1")
+  expect_equal(res$biosample, "SAMN03175161")
+})
+
+test_that("ncbi_link() assembly to biosample invalid queries", {
+  query <- c(examples$assembly, "noname", NA)
+  res <- ncbi_link(query, from = "assembly", to = "biosample")
+  
+  expect_equal(res$assembly, query)
+  expect_equal(res$assembly[4:5], c("noname", NA_character_))
+  expect_equal(res$biosample[4:5], c(NA_character_, NA_character_))
+})
+
+test_that("ncbi_link() biosample to assembly", {
+  query <- c(examples$biosample)
+  res <- ncbi_link(query, from = "biosample", to = "assembly")
+  
+  expect_equal(res$biosample, query)
+  expect_equal(
+    res$assembly[which(res$biosample == "SAMN02714232")],
+    "GCF_000695855.3"
+  )
+  expect_equal(
+    res$assembly[which(res$biosample == "SAMN36356470")],
+    NA_character_
+  )
+})