chore: Phase 2.8 grooming — couple CLI data_types to _DATATYPE_TO_PROFILE, reject inverted BED intervals, guard wgs+capture_bed at API level, drop unused fixture

Four fixes flagged by final cumulative review:
- CLI `valid_data_types` now derived from `_DATATYPE_TO_PROFILE.keys()`
  to prevent future drift between CLI guard and API mapping.
- `load_capture_bed` raises `CaptureBedError` on inverted intervals
  (start > end) rather than silently accepting rows that match zero
  positions — quiet correctness failure prevented.
- `Annotator.__init__` raises `ValueError` on `data_type='wgs'` with
  a `capture_bed` path, matching the CLI's existing rejection.
- Unused `sma_panel_bed_empty` fixture removed (YAGNI).

Two new tests cover Fix 2 and Fix 3. Test count 159 → 161.

Author: Enes Ak

src/locusguard/api.py

@@ -74,6 +74,10 @@ def __init__(
         self._reference_fasta = reference_fasta
         self._data_type = data_type
         self._capture_bed = capture_bed
+        if data_type == "wgs" and capture_bed is not None:
+            raise ValueError(
+                "capture_bed is only valid with data_type 'wes' or 'panel' — got 'wgs'"
+            )
         self._profile_name = _DATATYPE_TO_PROFILE.get(data_type)
 
     def annotate_vcf(

src/locusguard/capture_bed.py

@@ -84,6 +84,11 @@ def load_capture_bed(path: Path) -> list[CaptureRegion]:
                     f"Line {line_num}: start/end must be integers,"
                     f" got '{parts[1]}' / '{parts[2]}'"
                 ) from e
+            if start > end:
+                raise CaptureBedError(
+                    f"Line {line_num}: start ({start}) > end ({end})"
+                    f" — BED intervals must be non-inverted"
+                )
             regions.append(CaptureRegion(chrom=chrom, start=start, end=end))
     return regions
 

src/locusguard/cli/annotate.py

@@ -8,7 +8,7 @@
 import typer
 from rich.console import Console
 
-from locusguard.api import Annotator
+from locusguard.api import _DATATYPE_TO_PROFILE, Annotator
 from locusguard.config import load_config
 from locusguard.config.schema import LocusConfig
 from locusguard.io.reference import ReferenceNotFoundError, resolve_reference_fasta
@@ -68,10 +68,9 @@ def annotate(
     LOCUS_STATUS, LOCUS_EVIDENCE, LOCUS_KEY fields on variants inside configured
     locus regions.
     """
-    valid_data_types = {"wgs", "wes", "panel"}
-    if data_type not in valid_data_types:
+    if data_type not in _DATATYPE_TO_PROFILE:
         raise typer.BadParameter(
-            f"--data-type must be one of: {', '.join(sorted(valid_data_types))}"
+            f"--data-type must be one of: {', '.join(sorted(_DATATYPE_TO_PROFILE))}"
         )
     if data_type == "wgs" and capture_bed is not None:
         raise typer.BadParameter(

tests/conftest.py

@@ -318,11 +318,3 @@ def sma_panel_bed_missing_psv(tmp_path: Path) -> Path:
     bed = tmp_path / "panel_missing.bed"
     bed.write_text("chr5\t12000\t13500\n")
     return bed
-
-
-@pytest.fixture
-def sma_panel_bed_empty(tmp_path: Path) -> Path:
-    """Synthetic BED with zero regions (just track/comment lines)."""
-    bed = tmp_path / "panel_empty.bed"
-    bed.write_text("track name=empty_panel\n# nothing to see here\n")
-    return bed

tests/unit/test_api.py

@@ -127,3 +127,20 @@ def test_annotator_capture_bed_defaults_to_none():
         data_type="wgs",
     )
     assert annotator._capture_bed is None
+
+
+def test_annotator_rejects_wgs_with_capture_bed():
+    """Annotator API rejects data_type='wgs' + capture_bed combination
+    (matches the CLI's behavior for the library-level API)."""
+    from pathlib import Path
+
+    import pytest
+
+    from locusguard.api import Annotator
+    with pytest.raises(ValueError, match="capture_bed is only valid"):
+        Annotator(
+            configs=[],
+            reference_fasta=Path("/fake/ref.fa"),
+            data_type="wgs",
+            capture_bed=Path("/fake/cap.bed"),
+        )

tests/unit/test_capture_bed.py

@@ -180,3 +180,11 @@ def test_compute_empty_regions() -> None:
     cfg = _make_locus_config([("psv1", 150)])
     cov = compute_psv_coverage(cfg, [])
     assert cov == PsvCoverage(covered=[], missing=["psv1"], fraction_covered=0.0)
+
+
+def test_load_inverted_interval_raises(tmp_path: Path) -> None:
+    """start > end is rejected as a malformed BED row."""
+    p = tmp_path / "t.bed"
+    p.write_text("chr5\t5000\t100\n")
+    with pytest.raises(CaptureBedError, match="start .* > end"):
+        load_capture_bed(p)