pheval-exomiser 0.2.0__py3-none-any.whl → 0.2.2__py3-none-any.whl

pheval_exomiser/post_process/post_process_results_format.py

@@ -61,6 +61,7 @@ class PhEvalGeneResultFromExomiserJsonCreator:
 
 
 class PhEvalVariantResultFromExomiserJsonCreator:
+
     def __init__(self, exomiser_json_result: [dict], score_name: str):
         self.exomiser_json_result = exomiser_json_result
         self.score_name = score_name
@@ -97,7 +98,27 @@ class PhEvalVariantResultFromExomiserJsonCreator:
         """Return score from Exomiser result entry."""
         return round(result_entry[self.score_name], 4)
 
-    def extract_pheval_variant_requirements(self) -> [PhEvalVariantResult]:
+    def _filter_for_acmg_assignments(
+        self, variant: PhEvalVariantResult, score: float, variant_acmg_assignments: dict
+    ) -> bool:
+        """Filter variants if they meet the PATHOGENIC or LIKELY_PATHOGENIC ACMG classification."""
+        for assignment in variant_acmg_assignments:
+            if variant == PhEvalVariantResult(
+                chromosome=self._find_chromosome(assignment["variantEvaluation"]),
+                start=self._find_start_pos(assignment["variantEvaluation"]),
+                end=self._find_end_pos(assignment["variantEvaluation"]),
+                ref=self._find_ref(assignment["variantEvaluation"]),
+                alt=self._find_alt(assignment["variantEvaluation"]),
+                score=score,
+            ) and (
+                assignment["acmgClassification"] == "PATHOGENIC"
+                or assignment["acmgClassification"] == "LIKELY_PATHOGENIC"
+            ):
+                return True
+
+    def extract_pheval_variant_requirements(
+        self, use_acmg_filter: bool = False
+    ) -> [PhEvalVariantResult]:
         """Extract data required to produce PhEval variant output."""
         simplified_exomiser_result = []
         for result_entry in self.exomiser_json_result:
@@ -105,17 +126,23 @@ class PhEvalVariantResultFromExomiserJsonCreator:
                 if self.score_name in result_entry:
                     if "contributingVariants" in gene_hit:
                         score = self._find_relevant_score(result_entry)
-                        for cv in gene_hit["contributingVariants"]:
-                            simplified_exomiser_result.append(
-                                PhEvalVariantResult(
-                                    chromosome=self._find_chromosome(cv),
-                                    start=self._find_start_pos(cv),
-                                    end=self._find_end_pos(cv),
-                                    ref=self._find_ref(cv),
-                                    alt=self._find_alt(cv),
-                                    score=score,
-                                )
+                        contributing_variants = gene_hit["contributingVariants"]
+                        variant_acmg_assignments = gene_hit["acmgAssignments"]
+                        for cv in contributing_variants:
+                            variant = PhEvalVariantResult(
+                                chromosome=self._find_chromosome(cv),
+                                start=self._find_start_pos(cv),
+                                end=self._find_end_pos(cv),
+                                ref=self._find_ref(cv),
+                                alt=self._find_alt(cv),
+                                score=score,
                             )
+                            if use_acmg_filter and self._filter_for_acmg_assignments(
+                                variant, score, variant_acmg_assignments
+                            ):
+                                simplified_exomiser_result.append(variant)
+                            if not use_acmg_filter:
+                                simplified_exomiser_result.append(variant)
         return simplified_exomiser_result
 
 
@@ -166,6 +193,7 @@ def create_standardised_results(
     variant_analysis: bool,
     gene_analysis: bool,
     disease_analysis: bool,
+    include_acmg: bool = False,
 ) -> None:
     """Write standardised gene/variant/disease results from default Exomiser json output."""
     for exomiser_json_result in files_with_suffix(results_dir, ".json"):
@@ -183,7 +211,7 @@ def create_standardised_results(
         if variant_analysis:
             pheval_variant_requirements = PhEvalVariantResultFromExomiserJsonCreator(
                 exomiser_result, score_name
-            ).extract_pheval_variant_requirements()
+            ).extract_pheval_variant_requirements(include_acmg)
             generate_pheval_result(
                 pheval_result=pheval_variant_requirements,
                 sort_order_str=sort_order,
@@ -255,6 +283,13 @@ def create_standardised_results(
     default=False,
     help="Specify whether to create PhEval disease results.",
 )
+@click.option(
+    "--include-acmg",
+    is_flag=True,
+    type=bool,
+    default=False,
+    help="Specify whether to include ACMG filter for PATHOGENIC or LIKELY_PATHOGENIC classifications.",
+)
 def post_process_exomiser_results(
     output_dir: Path,
     results_dir: Path,
@@ -263,6 +298,7 @@ def post_process_exomiser_results(
     gene_analysis: bool,
     variant_analysis: bool,
     disease_analysis: bool,
+    include_acmg: bool,
 ):
     """Post-process Exomiser json results into PhEval gene and variant outputs."""
     (
@@ -288,4 +324,5 @@ def post_process_exomiser_results(
         variant_analysis,
         gene_analysis,
         disease_analysis,
+        include_acmg,
     )

pheval_exomiser/prepare/create_batch_commands.py

@@ -7,7 +7,7 @@ from typing import Optional
 import click
 from phenopackets import Family, Phenopacket
 from pheval.prepare.custom_exceptions import MutuallyExclusiveOptionError
-from pheval.utils.file_utils import all_files, files_with_suffix, obtain_closest_file_name
+from pheval.utils.file_utils import all_files, files_with_suffix
 from pheval.utils.phenopacket_utils import PhenopacketUtil, phenopacket_reader
 
 from pheval_exomiser.constants import (
@@ -69,7 +69,9 @@ class CommandCreator:
             return (
                 self.output_options_file
                 if self.output_options_dir_files is None
-                else obtain_closest_file_name(self.phenopacket_path, self.output_options_dir_files)
+                else self.output_options_dir_files[0].parent.joinpath(
+                    self.phenopacket_path.stem + "-output_options.yml"
+                )
             )
 
     def add_phenotype_only_arguments(self) -> ExomiserCommandLineArguments:
@@ -182,6 +184,8 @@ class CommandsWriter:
                 + str(command_arguments.vcf_file)
                 + " --assembly "
                 + command_arguments.vcf_assembly
+                + " --output-filename "
+                + f"{command_arguments.sample.stem}-exomiser"
             )
         except IOError:
             print("Error writing ", self.file)

pheval_exomiser/prepare/tool_specific_configuration_options.py

@@ -11,8 +11,10 @@ class ApplicationProperties(BaseModel):
         cadd_version (str): Version of the CADD database
         hg19_data_version (str): Data version of the hg19 Exomiser data
         hg19_local_frequency_path (Path): The file name of the hg19 local frequency file
+        hg19_whitelist_path (Path): The file name of the hg19 whitelist.
         hg38_data_version (str): Data version of the hg38 Exomiser data
         hg38_local_frequency_path (Path): The file name of the hg38 local frequency file
+        hg38_whitelist_path (Path): The file name of the hg38 whitelist.
         phenotype_data_version (str): Data version of the Exomiser phenotype data
         cache_caffeine_spec (int): Cache limit
     """
@@ -21,8 +23,10 @@ class ApplicationProperties(BaseModel):
     cadd_version: str = Field(None)
     hg19_data_version: str = Field(None)
     hg19_local_frequency_path: Path = Field(None)
+    hg19_whitelist_path: Path = Field(None)
     hg38_data_version: str = Field(None)
     hg38_local_frequency_path: Path = Field(None)
+    hg38_whitelist_path: Path = Field(None)
     phenotype_data_version: str = Field(None)
     cache_type: str = Field(None)
     cache_caffeine_spec: int = Field(None)

pheval_exomiser/prepare/write_application_properties.py

@@ -144,16 +144,18 @@ class ExomiserConfigurationFileWriter:
 
     def write_hg19_white_list_path(self) -> None:
         """Write the hg19 whitelist path to application.properties file."""
-        if self.configurations.application_properties.hg19_data_version is not None:
+        if self.configurations.application_properties.hg19_whitelist_path is not None:
             self.application_properties.write(
-                "exomiser.hg19.variant-white-list-path=${exomiser.hg19.data-version}_hg19_clinvar_whitelist.tsv.gz\n"
+                f"exomiser.hg19.variant-white-list-path="
+                f"{self.configurations.application_properties.hg19_whitelist_path}\n"
             )
 
     def write_hg38_white_list_path(self) -> None:
         """Write the hg38 whitelist path to application.properties file."""
-        if self.configurations.application_properties.hg38_data_version is not None:
+        if self.configurations.application_properties.hg38_whitelist_path is not None:
             self.application_properties.write(
-                "exomiser.hg38.variant-white-list-path=${exomiser.hg38.data-version}_hg38_clinvar_whitelist.tsv.gz\n"
+                f"exomiser.hg38.variant-white-list-path="
+                f"{self.configurations.application_properties.hg38_whitelist_path}\n"
             )
 
     def write_cache_type(self):

{pheval_exomiser-0.2.0.dist-info → pheval_exomiser-0.2.2.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: pheval_exomiser
-Version: 0.2.0
+Version: 0.2.2
 Summary: 
 Author: Yasemin Bridges
 Author-email: y.bridges@qmul.ac.uk
@@ -58,8 +58,10 @@ tool_specific_configuration_options:
     cadd_version:
     hg19_data_version: 2302
     hg19_local_frequency_path: # name of hg19 local frequency file 
+    hg19_whitelist_path: 2302_hg19_clinvar_whitelist.tsv.gz # only required for Exomiser v13.3.0 and earlier, can be left blank for Exomiser v14.0.0 onwards.
     hg38_data_version: 2302
     hg38_local_frequency_path: # name of hg38 local frequency file 
+    hg38_whitelist_path:
     phenotype_data_version: 2302
     cache_type:
     cache_caffeine_spec:
@@ -75,6 +77,8 @@ The `exomiser_software_directory` points to the name of the Exomiser distributio
 
 The analysis configuration file (in this case: `preset-exome-analysis.yml`) should be located within the input directory.
 
+The whitelist paths for the hg19 and hg38 dbs need only be specified for Exomiser v13.3.0 and earlier (unless specifying your own whitelist), as Exomiser v14.0.0 now includes this in the db.
+
 If using optional databases, such as REMM/CADD/local frequency the optional data input should look like so in the input
 directory:
 

{pheval_exomiser-0.2.0.dist-info → pheval_exomiser-0.2.2.dist-info}/RECORD

@@ -3,16 +3,16 @@ pheval_exomiser/cli.py,sha256=0SR1-L2sREEkFRfUPwYwkbSaBsz_L_Sxq1S4c9LQLJg,350
 pheval_exomiser/constants.py,sha256=o_pLWF8kX74BqyTsAZa7twwSKzedLnpupCI90k_bMqY,517
 pheval_exomiser/post_process/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 pheval_exomiser/post_process/post_process.py,sha256=ZLIGPeADGZn08jFc152QraiJnYSADlL35GOwxkCQDwA,901
-pheval_exomiser/post_process/post_process_results_format.py,sha256=AYjU1V4b4_GTVuLvuqvqsygX7ewU3cxHKmrLW9-z1o0,10410
+pheval_exomiser/post_process/post_process_results_format.py,sha256=WVtCQv5uiFk-6xL6zKsB0VWR2L52kCji_1fsXLitX3o,12112
 pheval_exomiser/prepare/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-pheval_exomiser/prepare/create_batch_commands.py,sha256=e_yIC9mALgZ_kop9wwR9XNkx5a3NOSrTjkmUG0k6Qvc,16088
-pheval_exomiser/prepare/tool_specific_configuration_options.py,sha256=YHAobmTCtsk6vbipT3IPyB3vG6dG4PldPfyAly2StMg,2427
-pheval_exomiser/prepare/write_application_properties.py,sha256=uFBNj6AEt3K-gNMNvFYRltVwVHtI-BzSsjLHy52ZeoQ,8877
+pheval_exomiser/prepare/create_batch_commands.py,sha256=G2VJ6xPgajYcNovVO1iC2PKePbU97-BqjbBONiC-hy4,16224
+pheval_exomiser/prepare/tool_specific_configuration_options.py,sha256=i4aXdEVfAnA6uQD4ZOk_OO4nXAVIFQ1nN_aVul56DZg,2661
+pheval_exomiser/prepare/write_application_properties.py,sha256=KmG7GvkQo8AhnhRyqohTFvqjfhEhbcs78UYYoigxJ3w,8933
 pheval_exomiser/prepare/yaml_to_family_phenopacket.py,sha256=Hz77dHpVaRMV1fQWKmOCqCKJfmk_hdpZh_6o7hq9Sec,14452
 pheval_exomiser/run/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 pheval_exomiser/run/run.py,sha256=6RokrunCoouhP_c4xpGR3EDzK6nxx_iOgyG3q1PYZmk,6892
 pheval_exomiser/runner.py,sha256=LaWhC0F9LoPvP0Ie1sG2GkC8EG-tWjBBY_tFYmx6dxA,2548
-pheval_exomiser-0.2.0.dist-info/METADATA,sha256=F8y5fYes22EQpgGq32NqC2PEMk9zgAOOzqEgQj7KlBQ,6684
-pheval_exomiser-0.2.0.dist-info/WHEEL,sha256=FMvqSimYX_P7y0a7UY-_Mc83r5zkBZsCYPm7Lr0Bsq4,88
-pheval_exomiser-0.2.0.dist-info/entry_points.txt,sha256=lbZMu-x7ns8UrFveWSqEQ1UB5l33TbRMomqBUyGYIwI,131
-pheval_exomiser-0.2.0.dist-info/RECORD,,
+pheval_exomiser-0.2.2.dist-info/METADATA,sha256=rdejFLD-k0Ngc8lEtSCrU-QSoLkKiI7pGroWcat8rfA,7059
+pheval_exomiser-0.2.2.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
+pheval_exomiser-0.2.2.dist-info/entry_points.txt,sha256=lbZMu-x7ns8UrFveWSqEQ1UB5l33TbRMomqBUyGYIwI,131
+pheval_exomiser-0.2.2.dist-info/RECORD,,

{pheval_exomiser-0.2.0.dist-info → pheval_exomiser-0.2.2.dist-info}/WHEEL

@@ -1,4 +1,4 @@
 Wheel-Version: 1.0
-Generator: poetry-core 1.8.1
+Generator: poetry-core 1.9.0
 Root-Is-Purelib: true
 Tag: py3-none-any