Merge pull request #109 from benjeffery/check_overlap

Error on overlapping partitions

Author: jeromekelleher

bio2zarr/vcf.py

@@ -284,6 +284,21 @@ def scan_vcf(path, target_num_partitions):
         return metadata, vcf.raw_header
 
 
+def check_overlap(partitions):
+    for i in range(1, len(partitions)):
+        prev_partition = partitions[i - 1]
+        current_partition = partitions[i]
+        if (
+            prev_partition.region.contig == current_partition.region.contig
+            and prev_partition.region.end > current_partition.region.start
+        ):
+            raise ValueError(
+                f"Multiple VCFs have the region "
+                f"{prev_partition.region.contig}:{prev_partition.region.start}-"
+                f"{current_partition.region.end}"
+            )
+
+
 def scan_vcfs(paths, show_progress, target_num_partitions, worker_processes=1):
     logger.info(
         f"Scanning {len(paths)} VCFs attempting to split into {target_num_partitions} partitions."
@@ -331,6 +346,7 @@ def scan_vcfs(paths, show_progress, target_num_partitions, worker_processes=1):
     all_partitions.sort(
         key=lambda x: (contig_index_map[x.region.contig], x.region.start)
     )
+    check_overlap(all_partitions)
     icf_metadata.partitions = all_partitions
     logger.info(f"Scan complete, resulting in {len(all_partitions)} partitions.")
     return icf_metadata, header

tests/test_vcf.py

@@ -1,11 +1,10 @@
 import json
-
 import pytest
 import xarray.testing as xt
 import sgkit as sg
 import zarr
 
-from bio2zarr import vcf
+from bio2zarr import vcf, vcf_utils
 
 
 @pytest.fixture(scope="module")
@@ -297,3 +296,25 @@ def test_call_genotype_phased(self, schema):
             },
             "filters": [],
         }
+
+
+@pytest.mark.parametrize(
+    "regions",
+    [
+        # Overlapping partitions
+        [("1", 100, 200), ("1", 150, 250)],
+        # Overlap by one position
+        [("1", 100, 201), ("1", 200, 300)],
+        # Contained overlap
+        [("1", 100, 300), ("1", 150, 250)],
+        # Exactly equal
+        [("1", 100, 200), ("1", 100, 200)],
+    ],
+)
+def test_check_overlap(regions):
+    partitions = [
+        vcf.VcfPartition("", region=vcf_utils.Region(contig, start, end))
+        for contig, start, end in regions
+    ]
+    with pytest.raises(ValueError, match="Multiple VCFs have the region"):
+        vcf.check_overlap(partitions)