Merge pull request #20 from yaseminbridges/18-use-pheval-as-a-dependency

18 use pheval as a dependency

pyproject.toml

@@ -14,6 +14,7 @@ polars = "^0.17.5"
 oaklib = "^0.5.1"
 click = "^8.1.3"
 ontobio = "^2.8.5"
+pheval = {git = "https://github.com/monarch-initiative/pheval.git"}
 
 [tool.poetry.dev-dependencies]
 pytest = "^7.2.0"

src/phenotype2phenopacket/add/add_genes.py

@@ -2,16 +2,16 @@
 
 import polars as pl
 from phenopackets import Disease
-
-from phenotype2phenopacket.utils.gene_map_utils import (
+from pheval.utils.phenopacket_utils import (
     GeneIdentifierUpdater,
     create_gene_identifier_map,
     create_hgnc_dict,
+    phenopacket_reader,
 )
+
 from phenotype2phenopacket.utils.phenopacket_utils import (
     PhenopacketInterpretationExtender,
     PhenopacketUtil,
-    phenopacket_reader,
     write_phenopacket,
 )
 from phenotype2phenopacket.utils.utils import all_files
@@ -66,20 +66,17 @@ def add_genes(
         ) if phenopacket_with_genes is not None else None
 
 
-def add_genes_to_directory(
-    phenopacket_dir: Path, disease_pg: pl.DataFrame, hgnc_data_file: Path, output_dir: Path
-):
+def add_genes_to_directory(phenopacket_dir: Path, disease_pg: pl.DataFrame, output_dir: Path):
     """
     Add known gene-to-phenotype relationships to the interpretations of a directory of phenopackets.
 
     Args:
         phenopacket_dir (Path): Directory containing the phenopacket files.
         disease_pg (pl.DataFrame): DataFrame containing disease.pg entries.
-        hgnc_data_file (Path): File path to HGNC data file.
         output_dir (Path): Directory to store the updated phenopackets.
     """
-    hgnc_dict = create_hgnc_dict(hgnc_data_file)
-    identifier_map = create_gene_identifier_map(hgnc_data_file)
+    hgnc_dict = create_hgnc_dict()
+    identifier_map = create_gene_identifier_map()
     gene_identifier_updater = GeneIdentifierUpdater(
         gene_identifier="ensembl_id", hgnc_data=hgnc_dict, identifier_map=identifier_map
     )

src/phenotype2phenopacket/cli_add.py

@@ -21,13 +21,6 @@
     help="Path to disease.pg data file.",
     type=Path,
 )
-@click.option(
-    "--hgnc-data",
-    "-h",
-    required=True,
-    help="Path to hgnc_full_set data file.",
-    type=Path,
-)
 @click.option(
     "--output-dir",
     "-o",
@@ -38,7 +31,6 @@
 def add_genes_command(
     phenopacket_dir: Path,
     disease_pg: Path,
-    hgnc_data: Path,
     output_dir: Path,
 ):
     """
@@ -47,14 +39,12 @@ def add_genes_command(
     Args:
         phenopacket_dir (Path): Directory containing the phenopacket files.
         disease_pg (Path): Path to the disease.pg file.
-        hgnc_data (Path): Path to the HGNC data file.
         output_dir (Path): Directory to store the updated phenopackets.
     """
     output_dir.mkdir(exist_ok=True)
     disease_pg_df = read_disease_pg(disease_pg)
     add_genes_to_directory(
         phenopacket_dir,
         disease_pg_df,
-        hgnc_data,
         output_dir,
     )

src/phenotype2phenopacket/utils/phenopacket_utils.py

@@ -1,4 +1,3 @@
-import json
 import re
 import secrets
 import signal
@@ -9,15 +8,13 @@
 from typing import List, Union
 
 import polars as pl
-from google.protobuf.json_format import MessageToJson, Parse
 from google.protobuf.timestamp_pb2 import Timestamp
 from oaklib.implementations import ProntoImplementation
 from ontobio import Ontology
 from phenopackets import (
     Age,
     Diagnosis,
     Disease,
-    Family,
     GeneDescriptor,
     GenomicInterpretation,
     Individual,
@@ -29,8 +26,8 @@
     Resource,
     TimeElement,
 )
+from pheval.utils.phenopacket_utils import GeneIdentifierUpdater, create_json_message
 
-from phenotype2phenopacket.utils.gene_map_utils import GeneIdentifierUpdater
 from phenotype2phenopacket.utils.utils import is_float
 
 
@@ -119,25 +116,6 @@ class PhenopacketFile:
     phenopacket_path: Path
 
 
-def phenopacket_reader(file: Path):
-    """
-    Read a Phenopacket file and returns its contents as a Phenopacket or Family object
-
-    Args:
-        file (Path): Path to the Phenopacket file
-
-    Returns:
-        Union[Phenopacket, Family]: Contents of the Phenopacket file as a Phenopacket or Family object
-    """
-    file = open(file, "r")
-    phenopacket = json.load(file)
-    file.close()
-    if "proband" in phenopacket:
-        return Parse(json.dumps(phenopacket), Family())
-    else:
-        return Parse(json.dumps(phenopacket), Phenopacket())
-
-
 def create_phenopacket_file_name_from_disease(disease_name: str) -> Path:
     """
     Create a Phenopacket file name from the disease.
@@ -149,19 +127,6 @@ def create_phenopacket_file_name_from_disease(disease_name: str) -> Path:
     return Path(normalised_string.replace(" ", "_") + ".json")
 
 
-def create_json_message(phenopacket: Phenopacket) -> str:
-    """
-    Create a JSON message for writing to a file.
-
-    Args:
-        phenopacket (Phenopacket): The Phenopacket object to convert to JSON.
-
-    Returns:
-        str: A JSON-formatted string representation of the Phenopacket or Family object.
-    """
-    return MessageToJson(phenopacket)
-
-
 def write_phenopacket(phenopacket: Phenopacket, output_file: Path) -> None:
     """
     Write a Phenopacket object to a file in JSON format.
@@ -172,11 +137,14 @@ def write_phenopacket(phenopacket: Phenopacket, output_file: Path) -> None:
     """
     phenopacket_json = create_json_message(phenopacket)
     suffix = 1
-    while Path(
-        output_file.parents[0].joinpath(f"{output_file.stem}_patient_{suffix}.json")
-    ).is_file():
-        suffix += 1
-    output_file = output_file.parents[0].joinpath(f"{output_file.stem}_patient_{suffix}.json")
+    if "_patient_" not in output_file.stem:
+        while Path(
+            output_file.parents[0].joinpath(f"{output_file.stem}_patient_{suffix}.json")
+        ).is_file():
+            suffix += 1
+        output_file = output_file.parents[0].joinpath(f"{output_file.stem}_patient_{suffix}.json")
+    else:
+        pass
     with open(output_file, "w") as file:
         file.write(phenopacket_json)
     file.close()

tests/test_phenopacket_utils.py

@@ -19,13 +19,13 @@
     Resource,
     TimeElement,
 )
-from polars.testing import assert_frame_equal
-
-from phenotype2phenopacket.utils.gene_map_utils import (
+from pheval.utils.phenopacket_utils import (
     GeneIdentifierUpdater,
     create_gene_identifier_map,
     create_hgnc_dict,
 )
+from polars.testing import assert_frame_equal
+
 from phenotype2phenopacket.utils.phenopacket_utils import (
     OnsetTerm,
     PhenopacketInterpretationExtender,
@@ -1183,12 +1183,8 @@ def setUpClass(cls) -> None:
         }
         cls.gene_identifier_updater = GeneIdentifierUpdater(
             gene_identifier="ensembl_id",
-            hgnc_data=create_hgnc_dict(
-                Path("./src/phenotype2phenopacket/resources/hgnc_complete_set_2023-04-01.txt")
-            ),
-            identifier_map=create_gene_identifier_map(
-                Path("./src/phenotype2phenopacket/resources/hgnc_complete_set_2023-04-01.txt")
-            ),
+            hgnc_data=create_hgnc_dict(),
+            identifier_map=create_gene_identifier_map(),
         )
 
     def test_create_gene_genomic_interpretation(self):