Include Puerto Rico when filtering Nextstrain metadata

Adding Puerto Rico to the US state filter per the WIP variant nowcast hub guidelines https://github.com/reichlab/variant-nowcast-hub/pull/39/files#r1759244751
reichlab · Sep 16, 2024 · 86530eb · 86530eb
1 parent 4e1da08
commit 86530eb
Show file tree

Hide file tree

Showing 4 changed files with 4 additions and 4 deletions.
diff --git a/src/virus_clade_utils/util/sequence.py b/src/virus_clade_utils/util/sequence.py
@@ -116,7 +116,7 @@ def filter_covid_genome_metadata(metadata: pl.LazyFrame, cols: list = []) -> pl.
 
     # There are some other odd divisions in the data, but these are 50 states and DC
     states = [state.name for state in us.states.STATES]
-    states.append("Washington DC")
+    states.extend(["Washington DC", "Puerto Rico"])
 
     # Filter dataset and do some general tidying
     filtered_metadata = (

diff --git a/tests/data/test_metadata.tsv b/tests/data/test_metadata.tsv
@@ -27,4 +27,4 @@ ghi	ghi.4	i ❤️ wombats	2024-09-22	Homo sapiens	USA	Utah	BB	Cardassia	hummus
 jkl	jkl.1	i ❤️ wombats	2024-09-22	Homo sapiens	USA	Utah	CC	Bajor	hummus a tune
 mno	mno.1	i ❤️ wombats	2024-09-22	Homo sapiens	Canada	Mississippi	FF	Earth	hummus a tune
 mno	mno.1	i ❤️ wombats	2024-09-22	marmots	USA	Massachusetts	FF	Cardassia	hummus a tune
-mno	mno.1	i ❤️ wombats	2024-09-22	Homo sapiens	USA	Puerto Rico	FF	Bajor	hummus a tune
+mno	mno.1	i ❤️ wombats	2024-09-22	Homo sapiens	USA	Guam	FF	Bajor	hummus a tune
diff --git a/tests/unit/test_get_clade_list.py b/tests/unit/test_get_clade_list.py
@@ -20,7 +20,7 @@ def test_file_path() -> Path:
         (0.1, 3, 9, ["AA", "AA.ZZ", "BB", "CC", "DD", "EE", "FF"]),
         (0.3, 3, 9, ["AA", "AA.ZZ", "EE"]),
         (0.1, 2, 9, ["AA.ZZ", "AA", "BB", "CC", "DD", "EE", "FF"]),
-        (0.1, 1, 9, ["AA", "BB", "CC", "FF"]),
+        (0.1, 1, 9, ["AA", "BB", "CC", "DD", "FF"]),
         (0.3, 1, 9, ["AA"]),
         (0.1, 3, 4, ["AA", "AA.ZZ", "BB", "CC"]),
         (0.3, 3, 2, ["AA", "AA.ZZ"]),

diff --git a/tests/unit/util/test_sequence.py b/tests/unit/util/test_sequence.py
@@ -58,7 +58,7 @@ def test_filter_covid_genome_metadata():
         "date": ["2022-01-01", "2022-01-02", "2022-01-03", "2023-12-25", None, "2023-12-27"],
         "host": ["Homo sapiens", "Homo sapiens", "Homo sapiens", "Narwhals", "Homo sapiens", "Homo sapiens"],
         "country": ["USA", "Argentina", "USA", "USA", "USA", "USA"],
-        "division": ["Alaska", "Maine", "Puerto Rico", "Massachusetts", "Utah", "Pennsylvania"],
+        "division": ["Alaska", "Maine", "Guam", "Puerto Rico", "Utah", "Pennsylvania"],
         "clade_nextstrain": ["AAA", "BBB", "CCC", "DDD", "EEE", "FFF"],
         "location": ["Vulcan", "Reisa", "Bajor", "Deep Space 9", "Earth", "Cardassia"],
         "genbank_accession": ["A1", "A2", "B1", "B2", "C1", "C2"],