From 19e400916e4bdb97b66c32e8c5bae3ebcb8c084f Mon Sep 17 00:00:00 2001
From: jkgoodrich <33063077+jkgoodrich@users.noreply.github.com>
Date: Thu, 19 Dec 2024 09:55:39 -0700
Subject: [PATCH 1/4] Add function `parse_variant` to create a Struct with the
 locus and alleles from a variant string or contig, position, ref, and alt.

---
 gnomad/utils/parse.py     | 63 ++++++++++++++++++++++++++++++++
 tests/utils/test_parse.py | 75 +++++++++++++++++++++++++++++++++++++++
 2 files changed, 138 insertions(+)
 create mode 100644 gnomad/utils/parse.py
 create mode 100644 tests/utils/test_parse.py

diff --git a/gnomad/utils/parse.py b/gnomad/utils/parse.py
new file mode 100644
index 000000000..1a63a3a29
--- /dev/null
+++ b/gnomad/utils/parse.py
@@ -0,0 +1,63 @@
+"""This module contains utility functions for general parsing."""
+import logging
+from typing import Optional
+
+import hail as hl
+
+logging.basicConfig(
+    format="%(asctime)s (%(name)s %(lineno)s): %(message)s",
+    datefmt="%m/%d/%Y %I:%M:%S %p",
+)
+logger = logging.getLogger("parse_utils")
+logger.setLevel(logging.INFO)
+
+
+def parse_variant(
+    variant_str: Optional[str] = None,
+    contig: Optional[str] = None,
+    position: Optional[int] = None,
+    ref: Optional[str] = None,
+    alt: Optional[str] = None,
+    build: Optional[str] = None,
+) -> hl.expr.StructExpression:
+    """
+    Create a Struct with the locus and alleles from a variant string or contig, position, ref, and alt.
+
+    :param variant_str: Variant string in the format contig-position-ref-alt or
+        contig:position:ref:alt.
+    :param contig: Chromosome of the variant.
+    :param position: Variant position.
+    :param ref: Reference allele.
+    :param alt: Alternate allele.
+    :param build: Reference genome build. If not provided, will infer from the variant
+        string or contig. If 'chr' is present in the contig, will assume GRCh38,
+        otherwise GRCh37.
+    :return: Struct with the locus and alleles.
+    """
+    if not variant_str and not all([contig, position, ref, alt]):
+        raise ValueError(
+            "Either `variant_str` must be provided or all of `contig`, `position`, "
+            "`ref`, and `alt`."
+        )
+
+    if not build:
+        build = "GRCh37"
+        if (variant_str and variant_str.startswith("chr")) or (contig and contig.startswith("chr")):
+            build = "GRCh38"
+
+        logger.info("No build provided. Assuming build: %s", build)
+
+    try:
+        if variant_str and ":" not in variant_str:
+            contig, position, ref, alt = variant_str.split("-")
+        if all([contig, position, ref, alt]):
+            variant_str = f"{contig}:{position}:{ref}:{alt}"
+
+        return hl.parse_variant(variant_str, reference_genome=build)
+
+    except:
+        raise ValueError(
+            f"Invalid variant format: {variant_str}. Valid formats: \n"
+            f"  contig-position-ref-alt \n"
+            f"  contig:position:ref:alt"
+        )
diff --git a/tests/utils/test_parse.py b/tests/utils/test_parse.py
new file mode 100644
index 000000000..56c4311e9
--- /dev/null
+++ b/tests/utils/test_parse.py
@@ -0,0 +1,75 @@
+"""Tests for the parse utility module."""
+
+import pytest
+import hail as hl
+from gnomad.utils.parse import parse_variant
+
+class TestParseVariant:
+    """Test the parse_variant function."""
+
+    @pytest.mark.parametrize(
+        "variant_str, contig, position, ref, alt, build, expected",
+        [
+            ("1-1000-A-T", None, None, None, None, None, hl.Struct(locus=hl.Locus("1", 1000, "GRCh37"), alleles=["A", "T"])),
+            ("chr1-1000-A-T", None, None, None, None, None, hl.Struct(locus=hl.Locus("chr1", 1000, "GRCh38"), alleles=["A", "T"])),
+            (None, "1", 1000, "A", "T", None, hl.Struct(locus=hl.Locus("1", 1000, "GRCh37"), alleles=["A", "T"])),
+            (None, "chr1", 1000, "A", "T", None, hl.Struct(locus=hl.Locus("chr1", 1000, "GRCh38"), alleles=["A", "T"])),
+            (None, "1", 1000, "A", "T", "GRCh37", hl.Struct(locus=hl.Locus("1", 1000, "GRCh37"), alleles=["A", "T"])),
+            ("1:1000:A:T", None, None, None, None, None, hl.Struct(locus=hl.Locus("1", 1000, "GRCh37"), alleles=["A", "T"])),
+        ],
+    )
+    def test_parse_variant(
+        self,
+        variant_str: str,
+        contig: str,
+        position: int,
+        ref: str,
+        alt: str,
+        build: str,
+        expected: hl.expr.StructExpression,
+    ) -> None:
+        """
+        Test valid parameters for the `parse_variant` function.
+
+        :param variant_str: Variant string.
+        :param contig: Chromosome of the variant.
+        :param position: Variant position.
+        :param ref: Reference allele.
+        :param alt: Alternate allele.
+        :param build: Reference genome build.
+        :param expected: Expected result.
+        :return: None.
+        """
+        result = hl.eval(parse_variant(variant_str, contig, position, ref, alt, build))
+        assert result == expected
+
+    @pytest.mark.parametrize(
+        "variant_str, contig, position, ref, alt, build",
+        [
+            (None, None, None, None, None, None),
+            ("invalid_variant", None, None, None, None, None),
+            (None, "1", None, "A", "T", None),
+        ],
+    )
+    def test_parse_variant_invalid(
+        self,
+        variant_str: str,
+        contig: str,
+        position: int,
+        ref: str,
+        alt: str,
+        build: str,
+    ) -> None:
+        """
+        Test invalid parameters for the `parse_variant` function.
+
+        :param variant_str: Variant string.
+        :param contig: Chromosome of the variant.
+        :param position: Variant position.
+        :param ref: Reference allele.
+        :param alt: Alternate allele.
+        :param build: Reference genome build.
+        :return: None.
+        """
+        with pytest.raises(ValueError):
+            parse_variant(variant_str, contig, position, ref, alt, build)

From e2154d0286a6ac3b35c36721c806e71d886b3c00 Mon Sep 17 00:00:00 2001
From: jkgoodrich <33063077+jkgoodrich@users.noreply.github.com>
Date: Thu, 19 Dec 2024 10:06:02 -0700
Subject: [PATCH 2/4] Format

---
 gnomad/utils/parse.py     |  5 ++++-
 tests/utils/test_parse.py | 16 ++++++++++------
 2 files changed, 14 insertions(+), 7 deletions(-)

diff --git a/gnomad/utils/parse.py b/gnomad/utils/parse.py
index 1a63a3a29..6393f01ce 100644
--- a/gnomad/utils/parse.py
+++ b/gnomad/utils/parse.py
@@ -1,4 +1,5 @@
 """This module contains utility functions for general parsing."""
+
 import logging
 from typing import Optional
 
@@ -42,7 +43,9 @@ def parse_variant(
 
     if not build:
         build = "GRCh37"
-        if (variant_str and variant_str.startswith("chr")) or (contig and contig.startswith("chr")):
+        if (variant_str and variant_str.startswith("chr")) or (
+            contig and contig.startswith("chr")
+        ):
             build = "GRCh38"
 
         logger.info("No build provided. Assuming build: %s", build)
diff --git a/tests/utils/test_parse.py b/tests/utils/test_parse.py
index 56c4311e9..9c255ed11 100644
--- a/tests/utils/test_parse.py
+++ b/tests/utils/test_parse.py
@@ -4,18 +4,22 @@
 import hail as hl
 from gnomad.utils.parse import parse_variant
 
+
 class TestParseVariant:
     """Test the parse_variant function."""
 
+    grch37_out = hl.Struct(locus=hl.Locus("1", 1000, "GRCh37"), alleles=["A", "T"])
+    grch38_out = hl.Struct(locus=hl.Locus("chr1", 1000, "GRCh38"), alleles=["A", "T"])
+
     @pytest.mark.parametrize(
         "variant_str, contig, position, ref, alt, build, expected",
         [
-            ("1-1000-A-T", None, None, None, None, None, hl.Struct(locus=hl.Locus("1", 1000, "GRCh37"), alleles=["A", "T"])),
-            ("chr1-1000-A-T", None, None, None, None, None, hl.Struct(locus=hl.Locus("chr1", 1000, "GRCh38"), alleles=["A", "T"])),
-            (None, "1", 1000, "A", "T", None, hl.Struct(locus=hl.Locus("1", 1000, "GRCh37"), alleles=["A", "T"])),
-            (None, "chr1", 1000, "A", "T", None, hl.Struct(locus=hl.Locus("chr1", 1000, "GRCh38"), alleles=["A", "T"])),
-            (None, "1", 1000, "A", "T", "GRCh37", hl.Struct(locus=hl.Locus("1", 1000, "GRCh37"), alleles=["A", "T"])),
-            ("1:1000:A:T", None, None, None, None, None, hl.Struct(locus=hl.Locus("1", 1000, "GRCh37"), alleles=["A", "T"])),
+            ("1-1000-A-T", None, None, None, None, None, grch37_out),
+            ("chr1-1000-A-T", None, None, None, None, None, grch38_out),
+            (None, "1", 1000, "A", "T", None, grch37_out),
+            (None, "chr1", 1000, "A", "T", None, grch38_out),
+            (None, "1", 1000, "A", "T", "GRCh37", grch37_out),
+            ("1:1000:A:T", None, None, None, None, None, grch37_out),
         ],
     )
     def test_parse_variant(

From 1a4c8b38541298ceb6068898b7fe173428a91361 Mon Sep 17 00:00:00 2001
From: jkgoodrich <33063077+jkgoodrich@users.noreply.github.com>
Date: Thu, 19 Dec 2024 10:18:31 -0700
Subject: [PATCH 3/4] isort

---
 tests/utils/test_parse.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/tests/utils/test_parse.py b/tests/utils/test_parse.py
index 9c255ed11..710429a29 100644
--- a/tests/utils/test_parse.py
+++ b/tests/utils/test_parse.py
@@ -1,7 +1,8 @@
 """Tests for the parse utility module."""
 
-import pytest
 import hail as hl
+import pytest
+
 from gnomad.utils.parse import parse_variant
 
 

From 2e8b64311a4af3af6edd00096c1faf4948b5280a Mon Sep 17 00:00:00 2001
From: jkgoodrich <33063077+jkgoodrich@users.noreply.github.com>
Date: Thu, 19 Dec 2024 11:51:32 -0700
Subject: [PATCH 4/4] Use BaseException

---
 gnomad/utils/parse.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/gnomad/utils/parse.py b/gnomad/utils/parse.py
index 6393f01ce..1d65f51d7 100644
--- a/gnomad/utils/parse.py
+++ b/gnomad/utils/parse.py
@@ -58,7 +58,7 @@ def parse_variant(
 
         return hl.parse_variant(variant_str, reference_genome=build)
 
-    except:
+    except BaseException:
         raise ValueError(
             f"Invalid variant format: {variant_str}. Valid formats: \n"
             f"  contig-position-ref-alt \n"