loqusdb 2.6.9__py3-none-any.whl → 2.7.0__py3-none-any.whl

loqusdb/build_models/variant.py

@@ -139,19 +139,19 @@ def get_coords(variant):
     return coordinates
 
 
-def build_variant(variant, case_obj, case_id=None, gq_treshold=None, genome_build=None):
+def build_variant(variant, case_obj, case_id=None, gq_threshold=None, gq_qual=False, genome_build=None):
     """Return a Variant object
 
     Take a cyvcf2 formated variant line and return a models.Variant.
 
     If criterias are not fullfilled, eg. variant have no gt call or quality
-    is below gq treshold then return None.
+    is below gq threshold then return None.
 
     Args:
         variant(cyvcf2.Variant)
         case_obj(Case): We need the case object to check individuals sex
         case_id(str): The case id
-        gq_treshold(int): Genotype Quality treshold
+        gq_threshold(int): Genotype Quality threshold
 
     Return:
         formated_variant(models.Variant): A variant dictionary
@@ -188,8 +188,14 @@ def build_variant(variant, case_obj, case_id=None, gq_treshold=None, genome_buil
             ind_id = ind_obj["ind_id"]
             # Get the index position for the individual in the VCF
             ind_pos = ind_obj["ind_index"]
-            gq = int(variant.gt_quals[ind_pos])
-            if gq_treshold and gq < gq_treshold:
+
+            if gq_qual:
+                gq = int(variant.QUAL)
+
+            if not gq_qual:
+                gq = int(variant.gt_quals[ind_pos])
+
+            if gq_threshold and gq < gq_threshold:
                 continue
 
             genotype = GENOTYPE_MAP[variant.gt_types[ind_pos]]

loqusdb/commands/load.py

@@ -54,7 +54,8 @@ def validate_profile_threshold(ctx, param, value):
     help="Do not store case information on variants",
 )
 @click.option("--ensure-index", is_flag=True, help="Make sure that the indexes are in place")
-@click.option("--gq-treshold", default=20, show_default=True, help="Treshold to consider variant")
+@click.option("--gq-threshold", default=20, show_default=True, help="Threshold to consider variant")
+@click.option("--qual-gq", is_flag=True, default=False, show_default=True, help="Use QUAL tag instead of GQ value for quality filter")
 @click.option(
     "--max-window",
     "-m",
@@ -89,13 +90,14 @@ def load(
     family_file,
     family_type,
     skip_case_id,
-    gq_treshold,
+    gq_threshold,
     case_id,
     ensure_index,
     max_window,
     check_profile,
     hard_threshold,
     soft_threshold,
+    qual_gq
 ):
     """Load the variants of a case
 
@@ -136,7 +138,8 @@ def load(
             family_type=family_type,
             skip_case_id=skip_case_id,
             case_id=case_id,
-            gq_treshold=gq_treshold,
+            gq_threshold=gq_threshold,
+            qual_gq=qual_gq,
             max_window=max_window,
             profile_file=variant_profile_path,
             hard_threshold=hard_threshold,

loqusdb/commands/update.py

@@ -47,7 +47,7 @@ LOG = logging.getLogger(__name__)
     help="Do not store case information on variants",
 )
 @click.option("--ensure-index", is_flag=True, help="Make sure that the indexes are in place")
-@click.option("--gq-treshold", default=20, show_default=True, help="Treshold to consider variant")
+@click.option("--gq-threshold", default=20, show_default=True, help="Threshold to consider variant")
 @click.option(
     "--max-window",
     "-m",
@@ -63,7 +63,7 @@ def update(
     family_file,
     family_type,
     skip_case_id,
-    gq_treshold,
+    gq_threshold,
     case_id,
     ensure_index,
     max_window,
@@ -102,7 +102,7 @@ def update(
             family_type=family_type,
             skip_case_id=skip_case_id,
             case_id=case_id,
-            gq_treshold=gq_treshold,
+            gq_threshold=gq_threshold,
             max_window=max_window,
         )
     except (SyntaxError, CaseError, IOError, VcfError) as error:

loqusdb/utils/load.py

@@ -31,7 +31,8 @@ def load_database(
     family_file=None,
     family_type="ped",
     skip_case_id=False,
-    gq_treshold=None,
+    gq_threshold=None,
+    qual_gq=False,
     case_id=None,
     max_window=3000,
     profile_file=None,
@@ -48,7 +49,8 @@ def load_database(
           family_file(str): Path to family file
           family_type(str): Format of family file
           skip_case_id(bool): If no case information should be added to variants
-          gq_treshold(int): If only quality variants should be considered
+          gq_threshold(int): If only quality variants should be considered
+          qual_gq(bool): Use QUAL field instead of GQ format tag to gate quality
           case_id(str): If different case id than the one in family file should be used
           max_window(int): Specify the max size for sv windows
           check_profile(bool): Does profile check if True
@@ -87,13 +89,13 @@ def load_database(
             adapter, profiles, hard_threshold=hard_threshold, soft_threshold=soft_threshold
         )
 
-    # If a gq treshold is used the variants needs to have GQ
+    # If a gq threshold is used the variants needs to have GQ
     for _vcf_file in vcf_files:
         # Get a cyvcf2.VCF object
         vcf = get_vcf(_vcf_file)
 
-        if gq_treshold and not vcf.contains("GQ"):
-            LOG.warning("Set gq-treshold to 0 or add info to vcf {0}".format(_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")
 
     # Get a ped_parser.Family object from family file
@@ -143,7 +145,8 @@ def load_database(
                 vcf_obj=vcf_obj,
                 case_obj=case_obj,
                 skip_case_id=skip_case_id,
-                gq_treshold=gq_treshold,
+                gq_threshold=gq_threshold,
+                qual_gq=qual_gq,
                 max_window=max_window,
                 variant_type=variant_type,
                 genome_build=genome_build,
@@ -189,7 +192,8 @@ def load_variants(
     vcf_obj,
     case_obj,
     skip_case_id=False,
-    gq_treshold=None,
+    gq_threshold=None,
+    qual_gq=False,
     max_window=3000,
     variant_type="snv",
     genome_build=None,
@@ -202,7 +206,7 @@ def load_variants(
         nr_variants(int)
         skip_case_id (bool): whether to include the case id on variant level
                              or not
-        gq_treshold(int)
+        gq_threshold(int)
         max_window(int): Specify the max size for sv windows
         variant_type(str): 'sv' or 'snv'
 
@@ -222,7 +226,7 @@ def load_variants(
     with click.progressbar(vcf_obj, label="Inserting variants", length=nr_variants) as bar:
 
         variants = (
-            build_variant(variant, case_obj, case_id, gq_treshold, genome_build=genome_build)
+            build_variant(variant, case_obj, case_id, gq_threshold, qual_gq, genome_build=genome_build)
             for variant in bar
         )
 

loqusdb/utils/update.py

@@ -28,7 +28,7 @@ def update_database(
     family_file=None,
     family_type="ped",
     skip_case_id=False,
-    gq_treshold=None,
+    gq_threshold=None,
     case_id=None,
     max_window=3000,
 ):
@@ -41,7 +41,7 @@ def update_database(
           family_file(str): Path to family file
           family_type(str): Format of family file
           skip_case_id(bool): If no case information should be added to variants
-          gq_treshold(int): If only quality variants should be considered
+          gq_threshold(int): If only quality variants should be considered
           case_id(str): If different case id than the one in family file should be used
           max_window(int): Specify the max size for sv windows
 
@@ -67,14 +67,14 @@ def update_database(
         vcf_files.append(sv_file)
         sv_individuals = vcf_info["individuals"]
 
-    # If a gq treshold is used the variants needs to have GQ
+    # If a gq threshold is used the variants needs to have GQ
     for _vcf_file in vcf_files:
         # Get a cyvcf2.VCF object
         vcf = get_vcf(_vcf_file)
 
-        if gq_treshold:
+        if gq_threshold:
             if not vcf.contains("GQ"):
-                LOG.warning("Set gq-treshold to 0 or add info to vcf {0}".format(_vcf_file))
+                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")
 
     # Get a ped_parser.Family object from family file
@@ -127,7 +127,7 @@ def update_database(
                 vcf_obj=vcf_obj,
                 case_obj=case_obj,
                 skip_case_id=skip_case_id,
-                gq_treshold=gq_treshold,
+                gq_threshold=gq_threshold,
                 max_window=max_window,
                 variant_type=variant_type,
             )

{loqusdb-2.6.9.dist-info → loqusdb-2.7.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: loqusdb
-Version: 2.6.9
+Version: 2.7.0
 Summary: Store observations of vcf variants in a mongodb
 Home-page: https://github.com/moonso/loqusdb
 Author: Måns Magnusson
@@ -29,7 +29,6 @@ Requires-Dist: numpy (==1.21.4)
 Requires-Dist: coloredlogs (==14.0)
 Requires-Dist: pyyaml (==5.4.0)
 Requires-Dist: vcftoolbox (==1.5)
-Requires-Dist: pip (==21.3.1)
 Requires-Dist: setuptools (==59.2.0)
 Requires-Dist: mongo-adapter (>=0.3.3)
 Requires-Dist: ped-parser

{loqusdb-2.6.9.dist-info → loqusdb-2.7.0.dist-info}/RECORD

@@ -4,18 +4,18 @@ loqusdb/log.py,sha256=CDcrCjzs9ef-d5Wg8Q_41bCOZRM5j8PyP06kNcynTj0,1691
 loqusdb/build_models/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 loqusdb/build_models/case.py,sha256=P3sfQkI_fH8u5iqecYWhV866lcHz4upWkepaea5MMIw,4255
 loqusdb/build_models/profile_variant.py,sha256=TbSxfVjESstS_FgbkOW4NQwMQVeTyhn9oc9yPZmDhzI,1021
-loqusdb/build_models/variant.py,sha256=TxKOwWB1kIELo5LR6VcRC4kx-bZLnygwy3KfK7i4FCY,6552
+loqusdb/build_models/variant.py,sha256=JE3o_htRQjpwwUPxmP5HJ_0Ax8BwKp610uBIo3HTaVA,6671
 loqusdb/commands/__init__.py,sha256=BXAN3UADgqPrkGczzjlLO9GyyQ96dnLnP7n92JlYHgo,603
 loqusdb/commands/annotate.py,sha256=748kImopE5WbaO1nuv3WUgIqezWFSsi7SBeWhOz26-s,1384
 loqusdb/commands/cli.py,sha256=wJD5S1BoCxtRTAobd1QtmQpzlngJg-mt1nsyD92fDD4,3176
 loqusdb/commands/delete.py,sha256=R6ysHKSMw1mmL4ZbktoUIKzdzDLQ3314YPYhIy1myic,1979
 loqusdb/commands/export.py,sha256=0V3S3QU9LKlR13w3KCGfqFliTYiDRCgNwusA27AEvmE,3254
 loqusdb/commands/identity.py,sha256=hzbnvniKgSNEwSeYHsxdNvVlqu_vXeOSLlNFnDXTQjA,779
-loqusdb/commands/load.py,sha256=rNDgficyqtasZCuXrRECiKA27OYsaCk0D-Bq8NZRZD0,4226
+loqusdb/commands/load.py,sha256=cVDdY7meBfcv8nMEGsjAX6aE-SKDOceGqM2vAvXPhko,4407
 loqusdb/commands/load_profile.py,sha256=cflCbF9f77_HCH8xPnN8zSSocvIffRMnC2LPE0j7Xq8,3336
 loqusdb/commands/migrate.py,sha256=2C8YL-zVqnpnqg3JIyUr0rbVnb8-AGPVWNhicHnPKLo,667
 loqusdb/commands/restore.py,sha256=eqPX0yao0IAYS5SbjCdlsfSJRBbRByBLISUU2hTzqqs,1492
-loqusdb/commands/update.py,sha256=TbjbpegBE9N_yor2x1C_SziEx1AKHGk5qCnBLQZQsCA,3206
+loqusdb/commands/update.py,sha256=zz3wueaJVqJ1FKact-rpY2az__5oa1LnZKf7mgqNGPk,3211
 loqusdb/commands/view.py,sha256=zQag5kgvUFa8nW9OVte_qjit0n8wGLc3C3hwLOMGY6o,5111
 loqusdb/commands/wipe.py,sha256=WTOjyNooCUhtmZ6pdcPFa0PZrFc9E_pkLbnat_zP96M,553
 loqusdb/constants/__init__.py,sha256=r6y2TN8BqbKuh2Uyxq0trh-3A9xiWeStqWlvEPp-rSA,1645
@@ -40,15 +40,15 @@ loqusdb/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 loqusdb/utils/annotate.py,sha256=cPNWlhsv6yoe3lxNfa9DytO5eACuM_mOJJw_mglVMN0,2646
 loqusdb/utils/case.py,sha256=aeTvyACJTDjzl-aOjAZaUzFMLisgFKMfcoXSvNAZz4s,2168
 loqusdb/utils/delete.py,sha256=-ddBM_QXKzlUN6egEJggKzXX1P-WEdi92HgaD1DJRtg,4843
-loqusdb/utils/load.py,sha256=r71MkYowZZVF6M4l-03zZXlppIgFBlf2aZwdmGHYCLU,8065
+loqusdb/utils/load.py,sha256=rujUk89lL5jO35QfO8NKGJkjEJmN6PQxohKkxtpnw6Y,8252
 loqusdb/utils/migrate.py,sha256=9Q6kdIi9TpFVzDYptlEE8RqPPS5wyzfM3F8egzmmBBk,1113
 loqusdb/utils/profiling.py,sha256=3OizF7CpYvSl9kyl2g4KGJxbIRUqWfmfLxn3843XYDk,9164
-loqusdb/utils/update.py,sha256=o5yVMnIqAQOex8FaZe1Z1Z0Uxp8Vcr9cu5SEOIKIyk4,4426
+loqusdb/utils/update.py,sha256=1edJG-u24FgOSxyXAQEiyTG4IyK-Uo3lSIl5qyzcXsI,4433
 loqusdb/utils/variant.py,sha256=Lq5x9egVB-3rExBiRceF67TaL4Hp2gGoWMSRBEcnm4Q,2088
 loqusdb/utils/vcf.py,sha256=ybmrTBEPYa0FbUXo8ttlwATk13RnKjX9eIDbRDwCiVE,5175
-loqusdb-2.6.9.dist-info/LICENSE,sha256=urpFcJXw3elN9kV2fFutc-lXegjuu2lqP_GSy8_CAbs,1054
-loqusdb-2.6.9.dist-info/METADATA,sha256=m_ONy-h_R46VbvR_trZIsBk1Ja3kkgWaO-vquh8dp3Y,6056
-loqusdb-2.6.9.dist-info/WHEEL,sha256=G16H4A3IeoQmnOrYV4ueZGKSjhipXx8zc8nu9FGlvMA,92
-loqusdb-2.6.9.dist-info/entry_points.txt,sha256=39QklW01vy9ilBLcRPgCP18kN6oKXXLoOK50gZE7Jbs,59
-loqusdb-2.6.9.dist-info/top_level.txt,sha256=lRdRO6hqPhJEjFhfNsbCgVWMztvkYsjiOGK9DAL0UAI,8
-loqusdb-2.6.9.dist-info/RECORD,,
+loqusdb-2.7.0.dist-info/LICENSE,sha256=urpFcJXw3elN9kV2fFutc-lXegjuu2lqP_GSy8_CAbs,1054
+loqusdb-2.7.0.dist-info/METADATA,sha256=Sg2aINOps7cTErVgDl6xszkTG7GWTPrGdwFUQ9onR50,6026
+loqusdb-2.7.0.dist-info/WHEEL,sha256=pkctZYzUS4AYVn6dJ-7367OJZivF2e8RA9b_ZBjif18,92
+loqusdb-2.7.0.dist-info/entry_points.txt,sha256=39QklW01vy9ilBLcRPgCP18kN6oKXXLoOK50gZE7Jbs,59
+loqusdb-2.7.0.dist-info/top_level.txt,sha256=lRdRO6hqPhJEjFhfNsbCgVWMztvkYsjiOGK9DAL0UAI,8
+loqusdb-2.7.0.dist-info/RECORD,,

{loqusdb-2.6.9.dist-info → loqusdb-2.7.0.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: bdist_wheel (0.37.1)
+Generator: bdist_wheel (0.40.0)
 Root-Is-Purelib: true
 Tag: py3-none-any