bio2zarr 0.0.4__py3-none-any.whl → 0.0.6__py3-none-any.whl

bio2zarr/vcf_utils.py

@@ -1,14 +1,14 @@
-from typing import IO, Any, Dict, Optional, Sequence, Union
 import contextlib
-import struct
-import pathlib
 import gzip
-from dataclasses import dataclass
 import os
+import pathlib
+import struct
+from dataclasses import dataclass
+from typing import IO, Any, Dict, Optional, Sequence, Union
 
-import numpy as np
 import cyvcf2
 import humanfriendly
+import numpy as np
 
 from bio2zarr.typing import PathType
 
@@ -38,7 +38,8 @@ def read_bytes_as_value(f: IO[Any], fmt: str, nodata: Optional[Any] = None) -> A
     fmt : str
         A Python `struct` format string.
     nodata : Optional[Any], optional
-        The value to return in case there is no further data in the stream, by default None
+        The value to return in case there is no further data in the stream,
+        by default None
 
     Returns
     -------
@@ -277,7 +278,8 @@ class TabixIndex:
         # Create file offsets for each element in the linear index
         file_offsets = np.array([get_file_offset(vfp) for vfp in linear_index])
 
-        # Calculate corresponding contigs and positions or each element in the linear index
+        # Calculate corresponding contigs and positions or each element in
+        # the linear index
         contig_indexes = np.hstack(
             [np.full(len(li), i) for (i, li) in enumerate(linear_indexes)]
         )
@@ -433,6 +435,22 @@ class IndexedVcf(contextlib.AbstractContextManager):
             if var.POS >= start:
                 yield var
 
+    def _filter_empty(self, regions):
+        """
+        Return all regions in the specified list that have one or more records.
+
+        Sometimes with Tabix indexes these seem to crop up:
+
+        - https://github.com/sgkit-dev/bio2zarr/issues/45
+        - https://github.com/sgkit-dev/bio2zarr/issues/120
+        """
+        ret = []
+        for region in regions:
+            variants = self.variants(region)
+            if next(variants, None) is not None:
+                ret.append(region)
+        return ret
+
     def partition_into_regions(
         self,
         num_parts: Optional[int] = None,
@@ -509,4 +527,4 @@ class IndexedVcf(contextlib.AbstractContextManager):
             if self.index.record_counts[ri] > 0:
                 regions.append(Region(self.sequence_names[ri]))
 
-        return regions
+        return self._filter_empty(regions)

{bio2zarr-0.0.4.dist-info → bio2zarr-0.0.6.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: bio2zarr
-Version: 0.0.4
+Version: 0.0.6
 Summary: Convert bioinformatics data to Zarr
 Home-page: https://github.com/pystatgen/bio2zarr
 Author: sgkit Developers

bio2zarr-0.0.6.dist-info/RECORD

@@ -0,0 +1,16 @@
+bio2zarr/__init__.py,sha256=KiUGyya-9RHNcBldB8Lc1g3rP3CRjaL-5Olben0_6qA,49
+bio2zarr/__main__.py,sha256=hO4vV-kPFgsYq0NQwG2r-WkserPL27oqae_tUvNB7yE,527
+bio2zarr/_version.py,sha256=c6ZQWSJeXXzGZ3WoZWjkA-MiNkBFXMIRV9kZPo4MQ_M,411
+bio2zarr/cli.py,sha256=iQIT3c_mos97st9bTLiBuxLL70MbNOtlTegUti6q2-g,14397
+bio2zarr/core.py,sha256=E8MF5PUVulYtq5gaxaMWN7g5YeWPDueqlGCck3DJCuM,8516
+bio2zarr/plink.py,sha256=huXMlxQ5C3gPmOYCavA-QW7PzaV48I2lo80cQqHT1wY,6768
+bio2zarr/provenance.py,sha256=c_Z__QbWkLS0Rfa8D7LgEhtStng_zRMJX8comaDXIkw,142
+bio2zarr/typing.py,sha256=BYxhL16sKRoNxa6amf6AYxvt5Ke9qzv2np_kOT_zPJo,79
+bio2zarr/vcf.py,sha256=dJcaRCdlDlUxYE_Z3hQaqSy7EivfpihP78XSRMzlWbU,83330
+bio2zarr/vcf_utils.py,sha256=r3NQXxWK1SYU7CcwDzSWXdX5Q8Ixk7gdCTEiFPzfUAk,17307
+bio2zarr-0.0.6.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+bio2zarr-0.0.6.dist-info/METADATA,sha256=siwhDelRAqr4Zo4EAPqZ4c9SkMOsU9ixRtgALAROP9A,1077
+bio2zarr-0.0.6.dist-info/WHEEL,sha256=GJ7t_kWBFywbagK5eo9IoUwLW6oyOeTKmQ-9iHFVNxQ,92
+bio2zarr-0.0.6.dist-info/entry_points.txt,sha256=pklStOdATE5hHJm4qiIvmhHkcn21Si_XAu6MC7ieNrk,131
+bio2zarr-0.0.6.dist-info/top_level.txt,sha256=ouAvp3u9N25eKrQbN8BCDLPcWWQLhtlgdHKu8AtEj5Q,9
+bio2zarr-0.0.6.dist-info/RECORD,,

bio2zarr-0.0.4.dist-info/RECORD

@@ -1,16 +0,0 @@
-bio2zarr/__init__.py,sha256=yIJYx4GyKtOLOtODOX0kGCeGPYgQ-TBbsRdT1NwBpQQ,37
-bio2zarr/__main__.py,sha256=3cgaQ4x8YKXt-9xC2GLrHnS6UA38y1GXqttwZiBZJg4,525
-bio2zarr/_version.py,sha256=yBVOKdXLEcTVc7YV7ZPqRXhRDRt-pKrfXxcgHkgPY5g,411
-bio2zarr/cli.py,sha256=QE0DfoZHbBbxq9K_im9y4tJ49_Wss0zzavSjjz-85Xw,11484
-bio2zarr/core.py,sha256=tZb9exfFmuzbA8tUpPY8avSm9YvfH31-vUCTM4fpj78,8128
-bio2zarr/plink.py,sha256=llhfP-v44BVPvgCcwXktk0YrKaJSII63U_PTtpHlGtM,6755
-bio2zarr/provenance.py,sha256=c_Z__QbWkLS0Rfa8D7LgEhtStng_zRMJX8comaDXIkw,142
-bio2zarr/typing.py,sha256=wZ99Zzp5BD9Nqpd-S5bn38fSdPzfj6Z9IHPBfZqt9Gs,78
-bio2zarr/vcf.py,sha256=MEskVTDq4QntzoawPz0sfmInV0aPkIPLXXNv7GmVcmY,73870
-bio2zarr/vcf_utils.py,sha256=_kMZdpye15HGpniv8wwISw0L6NEEi54ZFaTcM83wLGs,16751
-bio2zarr-0.0.4.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-bio2zarr-0.0.4.dist-info/METADATA,sha256=DISckjzZ0b6FpBTfBvpmJmEe00SIdTHyB3UTsTR8rws,1077
-bio2zarr-0.0.4.dist-info/WHEEL,sha256=GJ7t_kWBFywbagK5eo9IoUwLW6oyOeTKmQ-9iHFVNxQ,92
-bio2zarr-0.0.4.dist-info/entry_points.txt,sha256=pklStOdATE5hHJm4qiIvmhHkcn21Si_XAu6MC7ieNrk,131
-bio2zarr-0.0.4.dist-info/top_level.txt,sha256=ouAvp3u9N25eKrQbN8BCDLPcWWQLhtlgdHKu8AtEj5Q,9
-bio2zarr-0.0.4.dist-info/RECORD,,