[GEN-2027] Add staging for mutation in cis (#597)

* add staging for mutation in cis

* add R tests for getting database to synid mapping table

* add more strict conditions

Author: rxu17

R/mergeCheck.R

@@ -9,8 +9,8 @@ parser$add_argument("--staging",
                     action = "store_true",
                     help = "Use staging files")
 args <- parser$parse_args()
-testing <- args$testing
-staging <- args$staging
+testing <- as.logical(args$testing)
+staging <- as.logical(args$staging)
 
 library(synapser)
 library(VariantAnnotation)
@@ -39,13 +39,10 @@ variant_limit = 100000
 tbl_size_limit = 500
 
 #testing = as.logical(args[1])
-if (testing) {
-  databaseSynIdMappingId = 'syn11600968'
-} else if (staging) {
-  databaseSynIdMappingId = 'syn12094210'
-} else {
-  databaseSynIdMappingId = 'syn10967259'
-}
+databaseSynIdMappingId <- get_database_to_synapse_mapping_synid(
+  testing=testing, staging=staging
+)
+
 databaseSynIdMapping = synTableQuery(sprintf('select * from %s', databaseSynIdMappingId),
                                      includeRowIdAndRowVersion = F)
 databaseSynIdMappingDf = synapser::as.data.frame(databaseSynIdMapping)

R/mergecheck_functions.R

@@ -1,3 +1,20 @@
+#' Gets the database to synapse mapping table's synapse id based on
+#' if running in production, testing or staging mode
+#' @param testing (logical) running in testing mode
+#' @param staging (logical) running in staging mode
+get_database_to_synapse_mapping_synid <- function(testing, staging){
+  if (testing && !staging) {
+    databaseSynIdMappingId = 'syn11600968'
+  } else if (staging && !testing) {
+    databaseSynIdMappingId = 'syn12094210'
+  } else if (staging && testing) {
+    stop("Mutation in cis only available in staging or testing mode not both")
+  } else {
+    databaseSynIdMappingId = 'syn10967259'
+  }
+  return(databaseSynIdMappingId)
+}
+
 # Update mutation in cis table
 uploadToTable <- function(tbl, databaseSynId, subSetSamples, centerMappingDf) {
   # Old samples

R/tests/testthat/test_mergecheck_functions.R

@@ -0,0 +1,35 @@
+# tests for mergecheck_functions.R
+
+source("../../mergecheck_functions.R")
+
+library(testthat)
+
+
+test_that("get_database_to_synapse_mapping_synid_gets_correct_synid_for_testing", {
+  result <- get_database_to_synapse_mapping_synid(
+    testing = TRUE, staging = FALSE
+  )
+  expect_equal(result, "syn11600968")
+})
+
+test_that("get_database_to_synapse_mapping_synid_gets_correct_synid_for_staging", {
+  result <- get_database_to_synapse_mapping_synid(
+    testing = FALSE, staging = TRUE
+  )
+  expect_equal(result, "syn12094210")
+})
+
+test_that("get_database_to_synapse_mapping_synid_gets_correct_synid_for_prod", {
+  result <- get_database_to_synapse_mapping_synid(
+    testing = FALSE, staging = FALSE
+  )
+  expect_equal(result, "syn10967259")
+})
+
+test_that("get_database_to_synapse_mapping_synid_throws_error", {
+  expect_error(get_database_to_synapse_mapping_synid(testing = TRUE, staging = TRUE), 
+    "Mutation in cis only available in staging or testing mode not both")
+})
+
+
+

genie/database_to_staging.py

@@ -579,11 +579,6 @@ def mutation_in_cis_filter(
 
     The mutation in cis script ONLY runs
     WHEN the skipMutationsInCis parameter is FALSE
-    AND
-    WHEN staging parameter is FALSE
-
-    This is because we don't have this set up for staging mode
-    yet.
 
     Args:
         syn: Synapse object
@@ -597,8 +592,8 @@ def mutation_in_cis_filter(
     Returns:
         pd.Series: Samples to remove
     """
-    if (not skipMutationsInCis) and (not staging):
-        command = get_mutation_in_cis_filter_script_cmd(test=test)
+    if not skipMutationsInCis:
+        command = get_mutation_in_cis_filter_script_cmd(test=test, staging=staging)
         # TODO: use subprocess.run instead
         subprocess.check_call(command)
         store_mutation_in_cis_files_to_staging(
@@ -616,13 +611,14 @@ def mutation_in_cis_filter(
     return (remove_samples, flagged_variants)
 
 
-def get_mutation_in_cis_filter_script_cmd(test: bool) -> str:
+def get_mutation_in_cis_filter_script_cmd(test: bool, staging: bool) -> str:
     """This function gets the mutation_in_cis_filter R script
-        command call based on whether we are running in test,
+        command call based on whether we are running in test, staging
         or production mode
 
     Args:
         test (bool): Testing parameter.
+        staging (bool): Staging parameter. Default is False.
 
     Returns:
         str: Full command call for the mergeCheck script
@@ -631,8 +627,16 @@ def get_mutation_in_cis_filter_script_cmd(test: bool) -> str:
         os.path.dirname(os.path.abspath(__file__)), "../R/mergeCheck.R"
     )
     command = ["Rscript", mergeCheck_script]
-    if test:
+    if test and not staging:
         command.append("--testing")
+    if staging and not test:
+        command.append("--staging")
+    elif test and staging:
+        raise ValueError(
+            "Mutation in cis only available in staging or testing mode not both"
+        )
+    else:
+        pass
     return command
 
 

tests/test_database_to_staging.py

@@ -488,11 +488,9 @@ def test_that_store_maf_in_bed_filtered_variants_has_expected_calls(syn):
     [
         (True, False, False),
         (False, False, False),
+        (False, True, False),
     ],
-    ids=[
-        "testing_mode",
-        "prod_mode",
-    ],
+    ids=["testing_mode", "prod_mode", "staging_mode"],
 )
 def test_that_mutation_in_cis_filter_has_expected_calls_when_mutations_in_cis_is_not_skipped(
     syn, test, staging, skip_mutations_in_cis
@@ -544,7 +542,7 @@ def test_that_mutation_in_cis_filter_has_expected_calls_when_mutations_in_cis_is
             staging=staging,
         )
         # check expected calls with expected params
-        patch_get_cmd.assert_called_once_with(test=test)
+        patch_get_cmd.assert_called_once_with(test=test, staging=staging)
         patch_check_call.assert_called_once_with("test_cmd")
         patch_store_mutation_in_cis_files.assert_called_once_with(
             syn=syn,
@@ -592,13 +590,11 @@ def test_that_mutation_in_cis_filter_has_expected_calls_when_mutations_in_cis_is
     [
         (True, False, True),
         (False, False, True),
-        (False, True, False),
         (False, True, True),
     ],
     ids=[
         "testing_mode_skip_mutations_in_cis",
         "prod_mode_skip_mutations_in_cis",
-        "staging_mode",
         "staging_mode_skip_mutations_in_cis",
     ],
 )
@@ -692,32 +688,54 @@ def test_that_mutation_in_cis_filter_has_expected_calls_when_mutations_in_cis_is
 
 
 @pytest.mark.parametrize(
-    "test, expected_cmd",
+    "test, staging, expected_cmd",
     [
         (
             True,
+            False,
             [
                 "Rscript",
                 "/home/test_dir/Genie/../R/mergeCheck.R",
                 "--testing",
             ],
         ),
         (
+            False,
             False,
             [
                 "Rscript",
                 "/home/test_dir/Genie/../R/mergeCheck.R",
             ],
         ),
+        (
+            False,
+            True,
+            ["Rscript", "/home/test_dir/Genie/../R/mergeCheck.R", "--staging"],
+        ),
     ],
-    ids=["testing_mode", "prod_mode"],
+    ids=["testing_mode", "prod_mode", "staging_mode"],
 )
-def test_that_get_mutation_in_cis_filter_cmd_returns_correct_cmd(test, expected_cmd):
+def test_that_get_mutation_in_cis_filter_cmd_returns_correct_cmd(
+    test, staging, expected_cmd
+):
     with patch.object(os.path, "dirname", return_value="/home/test_dir/Genie/"):
-        cmd = database_to_staging.get_mutation_in_cis_filter_script_cmd(test=test)
+        cmd = database_to_staging.get_mutation_in_cis_filter_script_cmd(
+            test=test, staging=staging
+        )
         assert cmd == expected_cmd
 
 
+def test_that_get_mutation_in_cis_filter_cmd_raises_value_error():
+    with patch.object(os.path, "dirname", return_value="/home/test_dir/Genie/"):
+        with pytest.raises(
+            ValueError,
+            match="Mutation in cis only available in staging or testing mode not both",
+        ):
+            database_to_staging.get_mutation_in_cis_filter_script_cmd(
+                test=True, staging=True
+            )
+
+
 @pytest.mark.parametrize(
     "flagged_df, expected_flagged_variants",
     [