loqusdb 2.7.16__tar.gz → 2.7.18__tar.gz

{loqusdb-2.7.16 → loqusdb-2.7.18}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: loqusdb
-Version: 2.7.16
+Version: 2.7.18
 Summary: A simple observation count database
 License: MIT
 Author: Your Name

{loqusdb-2.7.16 → loqusdb-2.7.18}/loqusdb/__init__.py

@@ -4,7 +4,7 @@ from pymongo import ASCENDING, IndexModel
 
 logger = logging.getLogger(__name__)
 
-__version__ = "2.7.16"
+__version__ = "2.7.18"
 
 INDEXES = {
     "variant": [

{loqusdb-2.7.16 → loqusdb-2.7.18}/loqusdb/build_models/variant.py

@@ -148,6 +148,7 @@ def build_variant(
     case_id: Optional[str] = None,
     gq_threshold: Optional[int] = None,
     gq_qual: Optional[bool] = False,
+    ignore_gq_if_unset: Optional[bool] = False,
     genome_build: Optional[str] = None,
 ) -> Variant:
     """Return a Variant object
@@ -164,6 +165,8 @@ def build_variant(
         case_id(str): The case id
         gq_threshold(int): Genotype Quality threshold
         gq_qual(bool): Use variant.QUAL for quality instead of GQ
+        ignore_gq_if_unset(bool): Ignore GQ threshold check for variants that do not have GQ or QUAL set.
+        genome_build(str): Genome build. Ex. GRCh37 or GRCh38
 
     Return:
         formated_variant(models.Variant): A variant dictionary
@@ -201,14 +204,15 @@ def build_variant(
             ind_pos = ind_obj["ind_index"]
 
             if gq_qual:
-                gq = 0
+                gq = -1
                 if variant.QUAL:
                     gq = int(variant.QUAL)
 
             if not gq_qual:
                 gq = int(variant.gt_quals[ind_pos])
 
-            if gq_threshold and gq < gq_threshold:
+            # When gq is missing in FORMAT cyvcf2 assigns a score of -1
+            if (gq_threshold and 0 <= gq < gq_threshold) or (gq == -1 and not ignore_gq_if_unset):
                 continue
 
             genotype = GENOTYPE_MAP[variant.gt_types[ind_pos]]

{loqusdb-2.7.16 → loqusdb-2.7.18}/loqusdb/commands/load.py

@@ -88,6 +88,20 @@ 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.option(
+    "--ignore-gq-if-unset",
+    is_flag=True,
+    default=False,
+    show_default=True,
+    help="Ignore GQ threshold if GQ (or the QUAL field for --qual-gq) is unset in VCF",
+)
 @click.pass_context
 def load(
     ctx,
@@ -104,6 +118,8 @@ def load(
     hard_threshold,
     soft_threshold,
     qual_gq,
+    snv_gq_only,
+    ignore_gq_if_unset,
 ):
     """Load the variants of a case
 
@@ -145,12 +161,14 @@ 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,
             hard_threshold=hard_threshold,
             soft_threshold=soft_threshold,
             genome_build=genome_build,
+            ignore_gq_if_unset=ignore_gq_if_unset,
         )
     except (SyntaxError, CaseError, IOError) as error:
         LOG.warning(error)

{loqusdb-2.7.16 → loqusdb-2.7.18}/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,
@@ -38,6 +39,7 @@ def load_database(
     hard_threshold=0.95,
     soft_threshold=0.9,
     genome_build=None,
+    ignore_gq_if_unset=False,
 ):
     """Load the database with a case and its variants
 
@@ -55,6 +57,8 @@ def load_database(
           check_profile(bool): Does profile check if True
           hard_threshold(float): Rejects load if hamming distance above this is found
           soft_threshold(float): Stores similar samples if hamming distance above this is found
+          genome_build(str): Store the genome version
+          ignore_gq_if_unset(str): Ignore the gq threhsold check for variants that do not have a GQ or QUAL set
 
     Returns:
           nr_inserted(int)
@@ -88,11 +92,13 @@ 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")
@@ -144,11 +150,12 @@ 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,
                 genome_build=genome_build,
+                ignore_gq_if_unset=ignore_gq_if_unset,
             )
         except Exception as err:
             # If something went wrong do a rollback
@@ -196,18 +203,22 @@ def load_variants(
     max_window=3000,
     variant_type="snv",
     genome_build=None,
+    ignore_gq_if_unset=False,
 ):
     """Load variants for a family into the database.
 
     Args:
         adapter (loqusdb.plugins.Adapter): initialized plugin
+        vcf_obj(cyvcf2.VCF): Iterable with cyvcf2.Variant
         case_obj(Case): dict with case information
-        nr_variants(int)
         skip_case_id (bool): whether to include the case id on variant level
                              or not
         gq_threshold(int)
+        qual_gq(bool): whether to use QUAL instead of GQ
         max_window(int): Specify the max size for sv windows
         variant_type(str): 'sv' or 'snv'
+        genome_build(str): Genome version. Ex. GRCH37
+        ignore_gq_if_unset (bool): whether to add entries that have missing GQ or QUAL field
 
     Returns:
         nr_inserted(int)
@@ -226,7 +237,13 @@ def load_variants(
 
         variants = (
             build_variant(
-                variant, case_obj, case_id, gq_threshold, qual_gq, genome_build=genome_build
+                variant,
+                case_obj,
+                case_id,
+                gq_threshold,
+                qual_gq,
+                ignore_gq_if_unset,
+                genome_build=genome_build,
             )
             for variant in bar
         )

{loqusdb-2.7.16 → loqusdb-2.7.18}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "loqusdb"
-version = "2.7.16"
+version = "2.7.18"
 description = "A simple observation count database"
 authors = ["Your Name <you@example.com>"]
 license = "MIT"