loqusdb 2.7.14__tar.gz → 2.7.17__tar.gz

{loqusdb-2.7.14 → loqusdb-2.7.17}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: loqusdb
-Version: 2.7.14
+Version: 2.7.17
 Summary: A simple observation count database
 License: MIT
 Author: Your Name

{loqusdb-2.7.14 → loqusdb-2.7.17}/loqusdb/__init__.py

@@ -4,7 +4,7 @@ from pymongo import ASCENDING, IndexModel
 
 logger = logging.getLogger(__name__)
 
-__version__ = "2.7.14"
+__version__ = "2.7.17"
 
 INDEXES = {
     "variant": [

{loqusdb-2.7.14 → loqusdb-2.7.17}/loqusdb/commands/export.py

@@ -51,7 +51,13 @@ def export(ctx, outfile, variant_type, freq):
     for chrom in existing_chromosomes:
         ordered_chromosomes.append(chrom)
 
-    nr_cases = adapter.case_count()
+    if variant_type == "snv":
+        nr_cases = adapter.nr_cases(snv_cases=True)
+    elif variant_type == "sv":
+        nr_cases = adapter.nr_cases(sv_cases=True)
+    else:
+        raise ValueError(f"Unknown variant_type: {variant_type}, expected 'snv' or 'sv'")
+
     LOG.info(f"Found {nr_cases} cases in database")
 
     head = HeaderParser()

{loqusdb-2.7.14 → loqusdb-2.7.17}/loqusdb/commands/load.py

@@ -88,6 +88,13 @@ def validate_profile_threshold(ctx, param, value):
     callback=validate_profile_threshold,
     help="profile hamming distance to store similar individuals (0-1)",
 )
+@click.option(
+    "--snv-gq-only",
+    is_flag=True,
+    default=False,
+    show_default=True,
+    help="Apply GQ threshold only to SNV variants",
+)
 @click.pass_context
 def load(
     ctx,
@@ -104,6 +111,7 @@ def load(
     hard_threshold,
     soft_threshold,
     qual_gq,
+    snv_gq_only,
 ):
     """Load the variants of a case
 
@@ -145,6 +153,7 @@ def load(
             skip_case_id=skip_case_id,
             case_id=case_id,
             gq_threshold=gq_threshold,
+            snv_gq_only=snv_gq_only,
             qual_gq=qual_gq,
             max_window=max_window,
             profile_file=variant_profile_path,

{loqusdb-2.7.14 → loqusdb-2.7.17}/loqusdb/utils/load.py

@@ -31,6 +31,7 @@ def load_database(
     family_type="ped",
     skip_case_id=False,
     gq_threshold=None,
+    snv_gq_only=False,
     qual_gq=False,
     case_id=None,
     max_window=3000,
@@ -88,11 +89,13 @@ def load_database(
             adapter, profiles, hard_threshold=hard_threshold, soft_threshold=soft_threshold
         )
 
-    # If a gq threshold is used the variants needs to have GQ
+    # If a gq threshold is used the variants need to have GQ (only SNVs if snv_gq_only)
     for _vcf_file in vcf_files:
-        # Get a cyvcf2.VCF object
-        vcf = get_vcf(_vcf_file)
+        is_sv = _vcf_file == sv_file
+        if snv_gq_only and is_sv:
+            continue  # skip GQ check for SV VCF
 
+        vcf = get_vcf(_vcf_file)
         if gq_threshold and not vcf.contains("GQ") and not qual_gq:
             LOG.warning("Set gq-threshold to 0 or add info to vcf {0}".format(_vcf_file))
             raise SyntaxError("GQ is not defined in vcf header")
@@ -144,7 +147,7 @@ def load_database(
                 vcf_obj=vcf_obj,
                 case_obj=case_obj,
                 skip_case_id=skip_case_id,
-                gq_threshold=gq_threshold,
+                gq_threshold=gq_threshold if not snv_gq_only or variant_type == "snv" else None,
                 qual_gq=qual_gq,
                 max_window=max_window,
                 variant_type=variant_type,

{loqusdb-2.7.14 → loqusdb-2.7.17}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "loqusdb"
-version = "2.7.14"
+version = "2.7.17"
 description = "A simple observation count database"
 authors = ["Your Name <you@example.com>"]
 license = "MIT"