From ee950d353d44c25eecf62cc2f50f33bdd77425ad Mon Sep 17 00:00:00 2001
From: SkyeAv <skye.a.goetz@gmail.com>
Date: Tue, 25 Aug 2026 10:19:46 -0700
Subject: [PATCH] fix: leave number_of_cases alone instead of coercing it to
 study_size

number_of_cases is a legitimate Biolink Association slot (cases carrying
the phenotype/disease), but STUDY_SIZE_PREFIX_PATTERN matched it via
'number of cases' and coerce_study_size_columns renamed it to
study_size, destroying the edge field. Add STUDY_SIZE_EXEMPT_PATTERN so
study_size_target returns None for the exact slot.
---
 src/tablassert/coerce.py | 14 ++++++++++++++
 tests/test_lib.py        | 16 ++++++++++++++++
 2 files changed, 30 insertions(+)

diff --git a/src/tablassert/coerce.py b/src/tablassert/coerce.py
index f25e8a2..44212f9 100644
--- a/src/tablassert/coerce.py
+++ b/src/tablassert/coerce.py
@@ -415,6 +415,18 @@ def coerce_pvalue_columns(lf: pl.LazyFrame) -> pl.LazyFrame:
     """,
     re.IGNORECASE | re.VERBOSE,
 )
+# The Biolink ``Association`` slot ``number_of_cases`` counts cases carrying the
+# phenotype/disease, not the study population. It would otherwise match
+# PREFIX ("number of cases") and be destroyed by the rename to ``study_size``,
+# so the exact slot (separator-tolerant, like the patterns above) is exempt.
+STUDY_SIZE_EXEMPT_PATTERN: re.Pattern[str] = re.compile(
+    rf"""
+    ^
+    number {_SEP} of {_SEP} cases
+    $
+    """,
+    re.IGNORECASE | re.VERBOSE,
+)
 
 
 def study_size_target(name: str) -> str | None:
@@ -430,6 +442,8 @@ def study_size_target(name: str) -> str | None:
         ``"study_size"`` when the name matches any of the study-size patterns,
         else ``None``.
     """
+    if STUDY_SIZE_EXEMPT_PATTERN.search(name):
+        return None
     if STUDY_SIZE_EXACT_PATTERN.search(name):
         return "study_size"
     if STUDY_SIZE_COUNT_PATTERN.search(name):
diff --git a/tests/test_lib.py b/tests/test_lib.py
index f32bfd2..73c1bdc 100644
--- a/tests/test_lib.py
+++ b/tests/test_lib.py
@@ -2105,6 +2105,22 @@ def test_study_size_target_excludes_expanded_near_misses() -> None:
         assert study_size_target(n) is None, n
 
 
+def test_study_size_target_leaves_number_of_cases_alone() -> None:
+    """study_size_target never touches the Biolink ``number_of_cases`` slot.
+
+    ``number_of_cases`` counts cases carrying the phenotype/disease, not the
+    study population; coercing it to ``study_size`` destroys a legitimate edge
+    field.
+    """
+    names: list[str] = ["number_of_cases", "Number of Cases", "number-of-cases", "numberofcases"]
+    for n in names:
+        assert study_size_target(n) is None, n
+
+    lf: pl.LazyFrame = pl.DataFrame({"number_of_cases": [42, 7]}).lazy()
+    result: pl.DataFrame = coerce_study_size_columns(lf).collect()
+    assert result.columns == ["number_of_cases"]
+
+
 def test_coerce_study_size_columns_renames_n_column() -> None:
     """coerce_study_size_columns renames bare N to study_size."""
     lf: pl.LazyFrame = pl.DataFrame({"n": [120, 450]}).lazy()