Added a skip_aggregate_stats parameter to Dataset.delete_samples() in the Python API

This uses the tiledb_vcf_writer_set_skip_aggregate_stats hook in the C API to skip updating allele count and variant stats when deleting samples. Python tests were added to cover this new parameter.

Author: alancleary

apis/python/src/tiledbvcf/binding/libtiledbvcf.cc

@@ -131,6 +131,7 @@ PYBIND11_MODULE(libtiledbvcf, m) {
           "ingest_samples",
           &Writer::ingest_samples,
           py::call_guard<py::gil_scoped_release>())
+      .def("set_skip_aggregate_stats", &Writer::set_skip_aggregate_stats)
       .def(
           "delete_samples",
           &Writer::delete_samples,

apis/python/src/tiledbvcf/binding/writer.cc

@@ -232,6 +232,11 @@ void Writer::ingest_samples() {
   check_error(writer, tiledb_vcf_writer_store(writer));
 }
 
+void Writer::set_skip_aggregate_stats(bool skip) {
+  auto writer = ptr.get();
+  check_error(writer, tiledb_vcf_writer_set_skip_aggregate_stats(writer, skip));
+}
+
 void Writer::delete_samples(std::vector<std::string> samples_to_delete) {
   std::vector<const char*> samples;
   for (std::string& sample : samples_to_delete) {

apis/python/src/tiledbvcf/binding/writer.h

@@ -162,6 +162,8 @@ class Writer {
 
   void ingest_samples();
 
+  void set_skip_aggregate_stats(bool skip);
+
   void delete_samples(std::vector<std::string> samples);
 
   /** Returns schema version number of the TileDB VCF dataset */

apis/python/src/tiledbvcf/dataset.py

@@ -1111,9 +1111,23 @@ def ingest_samples(
     def delete_samples(
         self,
         sample_uris: List[str] = None,
+        skip_aggregate_stats: bool = False,
     ):
+        """
+        Delete samples from the dataset.
+
+        Parameters
+        ----------
+        sample_uris
+            List of sample names to delete.
+        skip_aggregate_stats
+            If True, skip updating allele_count and variant_stats arrays
+            during deletion.
+        """
         if self.mode != "w":
             raise Exception("Dataset not open in write mode")
+        if skip_aggregate_stats:
+            self.writer.set_skip_aggregate_stats(True)
         self.writer.delete_samples(sample_uris)
 
     def tiledb_stats(self) -> str:

apis/python/tests/test_delete.py

@@ -89,6 +89,45 @@ def test_delete_samples_nonexistent_raises(tmp_path):
         ds.delete_samples(["NONEXISTENT"])
 
 
+@skip_if_no_bcftools
+def test_delete_samples_skip_aggregate_stats(tmp_path, stats_v3_dataset):
+    """Verify skip_aggregate_stats skips stats update and records metadata."""
+    uri = os.path.join(tmp_path, "stats_test")
+    ds = tiledbvcf.Dataset(uri=uri, mode="w")
+    ds.delete_samples(["second"], skip_aggregate_stats=True)
+
+    # Verify sample is deleted
+    ds = tiledbvcf.Dataset(uri=uri, mode="r")
+    sample_names = ds.samples()
+    assert "second" not in sample_names
+    assert "third" in sample_names
+
+    # Verify skipped_delete_samples metadata on allele_count and variant_stats
+    for array_name in ["allele_count", "variant_stats"]:
+        with tiledb.open(os.path.join(uri, array_name), "r") as arr:
+            meta = arr.meta["skipped_delete_samples"]
+            assert b"second" in meta
+
+
+@skip_if_no_bcftools
+def test_delete_samples_skip_aggregate_stats_accumulates(
+    tmp_path, stats_v3_dataset
+):
+    """Verify skipped_delete_samples metadata accumulates across deletions."""
+    uri = os.path.join(tmp_path, "stats_test")
+
+    ds = tiledbvcf.Dataset(uri=uri, mode="w")
+    ds.delete_samples(["second"], skip_aggregate_stats=True)
+
+    ds = tiledbvcf.Dataset(uri=uri, mode="w")
+    ds.delete_samples(["fifth"], skip_aggregate_stats=True)
+
+    for array_name in ["allele_count", "variant_stats"]:
+        with tiledb.open(os.path.join(uri, array_name), "r") as arr:
+            meta = arr.meta["skipped_delete_samples"]
+            assert meta == b"second,fifth"
+
+
 def test_delete_samples_read_mode_raises(tmp_path):
     """Verify delete_samples() raises in read mode."""
     uri = os.path.join(tmp_path, "dataset")