Merge pull request #54 from jeromekelleher/multi-chr

Multi chr

Author: jeromekelleher

bio2zarr/core.py

@@ -225,8 +225,10 @@ def results_as_completed(self):
     def __exit__(self, exc_type, exc_val, exc_tb):
         if exc_type is None:
             wait_on_futures(self.futures)
+            # Note: this doesn't seem to be working correctly. If
+            # we set a timeout of None we get deadlocks
             set_progress(self.progress_config.total)
-            timeout = None
+            timeout = 1
         else:
             cancel_futures(self.futures)
             timeout = 0

bio2zarr/vcf.py

@@ -199,7 +199,6 @@ def make_field_def(name, vcf_type, vcf_number):
 
 
 def scan_vcf(path, target_num_partitions):
-    logger.debug(f"Scanning {path}")
     with vcf_utils.IndexedVcf(path) as indexed_vcf:
         vcf = indexed_vcf.vcf
         filters = [
@@ -238,6 +237,8 @@ def scan_vcf(path, target_num_partitions):
             pass
 
         regions = indexed_vcf.partition_into_regions(num_parts=target_num_partitions)
+        logger.info(
+            f"Split {path} into {len(regions)} regions (target={target_num_partitions})")
         for region in regions:
             metadata.partitions.append(
                 VcfPartition(

bio2zarr/vcf_utils.py

@@ -369,13 +369,16 @@ class IndexedVcf(contextlib.AbstractContextManager):
     def __init__(self, vcf_path, index_path=None):
         self.vcf = None
         vcf_path = pathlib.Path(vcf_path)
+        if not vcf_path.exists():
+            raise FileNotFoundError(vcf_path)
         # TODO use constants here instead of strings
         if index_path is None:
             index_path = vcf_path.with_suffix(vcf_path.suffix + ".tbi")
             if not index_path.exists():
                 index_path = vcf_path.with_suffix(vcf_path.suffix + ".csi")
                 if not index_path.exists():
-                    raise ValueError("Cannot find .tbi or .csi file.")
+                    raise FileNotFoundError(
+                        "Cannot find .tbi or .csi file for {vcf_path}")
         else:
             index_path = pathlib.Path(index_path)