address comments

utils/extract-pdbids-drugbank-xsdata-plugin/Dockerfile

@@ -14,14 +14,12 @@ COPY README.md ${EXEC_DIR}
 COPY CHANGELOG.md ${EXEC_DIR}
 
 # Install needed packages here
-# errors installing rdkit from poetry so using conda
-COPY environment.yml ${EXEC_DIR}
-RUN mamba env create -f ${EXEC_DIR}/environment.yml
-RUN echo "source activate project_env" > ~/.bashrc
-ENV PATH /opt/conda/envs/env/bin:$PATH
 
 COPY src ${EXEC_DIR}/src
 
+ADD Dockerfile .
+
 RUN pip3 install ${EXEC_DIR} --no-cache-dir
+#RUN pip3 install -i https://test.pypi.org/simple/ drugbank-schemas
 
-CMD ["--help"]
+#CMD ["--help"]

utils/extract-pdbids-drugbank-xsdata-plugin/build-docker.sh

@@ -1,4 +1,4 @@
 #!/bin/bash
 
 version=$(<VERSION)
-docker build . -t polusai/extract-pdbids-drugbank-xsdata-plugin:${version}
+docker build . -t polusai/extract-pdbids-drugbank-xsdata-tool:${version}

utils/extract-pdbids-drugbank-xsdata-plugin/extract_pdbids_drugbank_xsdata.cwl → utils/extract-pdbids-drugbank-xsdata-plugin/extract_pdbids_drugbank_xsdata_0@1@0.cwl

@@ -8,19 +8,18 @@ label: Filter Drugbank database using xsData
 doc: |-
   Filter Drugbank database using xsData
 
-baseCommand: ["conda", "run", "-n", "project_env", "python", "-m", "polus.mm.utils.extract_pdbids_drugbank_xsdata"]
+baseCommand: ["python", "-m", "polus.mm.utils.extract_pdbids_drugbank_xsdata"]
 
 hints:
   DockerRequirement:
-    dockerPull: ndonyapour/extract_pdbids_drugbank_xsdata
+    dockerPull: ndonyapour/extract_pdbids_drugbank_xsdata_tool
 
 requirements:
   InlineJavascriptRequirement: {}
   # Enabling InitialWorkDirRequirement will stage the input Drugbank xml file
   InitialWorkDirRequirement:
     listing:
     - $(inputs.drugbank_xml_file_path)
-    - $(inputs.drugbank_xsd_file_path)
 
 inputs:
   drugbank_xml_file_path:
@@ -31,21 +30,6 @@ inputs:
     format: edam:format_2332
     inputBinding:
       prefix: --drugbank_xml_file_path
-    default:
-      class: File
-      location: ../../../drugbank/drugbank_5.1.10.xml
-
-  drugbank_xsd_file_path:
-    label: Path to the Drugbank schema XSD file
-    doc: |-
-      Path to the Drugbank schema XSD file
-    type: File
-    format: edam:format_3804
-    inputBinding:
-      prefix: --drugbank_xsd_file_path
-    default:
-      class: File
-      location: ../../drugbank/drugbank_5.1.10.xsd
 
   smiles:
     label: List of input SMILES  # type:

utils/extract-pdbids-drugbank-xsdata-plugin/pyproject.toml

@@ -1,15 +1,30 @@
 [tool.poetry]
 name = "polus-mm-utils-extract-pdbids-drugbank-xsdata"
 version = "0.1.0"
-description = "Filter Drugbank database using xsData"
+description = "Filter DrugBank database using xsData"
 authors = ["Nazanin Donyapour <[email protected]>", "Brandon Walker <[email protected]>"]
 readme = "README.md"
 packages = [{include = "polus", from = "src"}]
 
 [tool.poetry.dependencies]
-python = ">=3.9,<3.12"
+python = ">=3.9,<3.13"
 typer = "^0.7.0"
-cwltool = "3.1.20240404144621"
+sophios = "0.1.4"
+pandas = "2.2.2"
+rdkit = "2024.3.5"
+
+# Specifying xsdata-pydantic with extra options
+xsdata-pydantic = { version = "*", extras = ["cli", "lxml", "soap"] }
+
+
+# Specifying additional index and package
+drugbank-schemas = { version = "0.1.4", source = "test-pypi"}
+
+# Ensure the correct priority for the additional index
+[[tool.poetry.source]]
+name = "test-pypi"
+url = "https://test.pypi.org/simple/"
+priority = "explicit"
 
 [tool.poetry.group.dev.dependencies]
 bump2version = "^1.0.1"
@@ -19,18 +34,10 @@ pre-commit = "^3.2.1"
 black = "^23.3.0"
 mypy = "^1.1.1"
 ruff = "^0.3.0"
-drugbank_schemas = { version = "0.1.4", source = "testpypi" }
-
-[[tool.poetry.source]]
-name = "testpypi"
-url = "https://test.pypi.org/simple/"
-secondary = true
 
 [build-system]
 requires = ["poetry-core"]
 build-backend = "poetry.core.masonry.api"
 
 [tool.pytest.ini_options]
-pythonpath = [
-  "."
-]
+pythonpath = ["."]

utils/extract-pdbids-drugbank-xsdata-plugin/src/polus/mm/utils/extract_pdbids_drugbank_xsdata/__main__.py

@@ -1,10 +1,10 @@
 """Package entrypoint for the extract_pdbids_drugbank_xsdata package."""
 
 # Base packages
+import argparse
 import logging
 from os import environ
 
-import typer
 from polus.mm.utils.extract_pdbids_drugbank_xsdata.extract_pdbids_drugbank_xsdata import (  # noqa: E501
     extract_pdbids_drugbank_xsdata,
 )
@@ -17,52 +17,60 @@
 logger = logging.getLogger("polus.mm.utils.extract_pdbids_drugbank_xsdata.")
 logger.setLevel(POLUS_LOG)
 
-app = typer.Typer(help="extract_pdbids_drugbank_xsdata.")
+
+def main(args: argparse.Namespace) -> None:
+    """extract_pdbids_drugbank_xsdata."""
+    logger.info(f"drugbank_xml_file_path: {args.drugbank_xml_file_path}")
+    logger.info(f"smiles: {args.smiles}")
+    logger.info(f"inchi: {args.inchi}")
+    logger.info(f"inchi_keys: {args.inchi_keys}")
+    logger.info(f"output_txt_path: {args.output_txt_path}")
+
+    extract_pdbids_drugbank_xsdata(
+        drugbank_xml_file_path=args.drugbank_xml_file_path,
+        smiles=args.smiles,
+        inchi=args.inchi,
+        inchi_keys=args.inchi_keys,
+        output_txt_path=args.output_txt_path,
+    )
 
 
-@app.command()
-def main(
-    drugbank_xml_file_path: str = typer.Option(
-        ...,
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(description="extract_pdbids_drugbank_xsdata.")
+
+    parser.add_argument(
         "--drugbank_xml_file_path",
+        type=str,
+        required=True,
         help="Path to the Drugbank xml file",
-    ),
-    smiles: list[str] = typer.Option(
-        ...,
+    )
+    parser.add_argument(
         "--smiles",
+        type=str,
+        nargs="+",
+        required=True,
         help="List of input SMILES, Type string[], File type input",
-    ),
-    inchi: list[str] = typer.Option(
-        ...,
+    )
+    parser.add_argument(
         "--inchi",
-        help="List of input SMILES, Type string[], File type input",
-    ),
-    inchi_keys: list[str] = typer.Option(
-        ...,
+        type=str,
+        nargs="+",
+        required=True,
+        help="List of input InChI, Type string[], File type input",
+    )
+    parser.add_argument(
         "--inchi_keys",
-        help="List of input SMILES, Type string[], File type input",
-    ),
-    output_txt_path: str = typer.Option(
-        ...,
+        type=str,
+        nargs="+",
+        required=True,
+        help="List of input InChI keys, Type string[], File type input",
+    )
+    parser.add_argument(
         "--output_txt_path",
+        type=str,
+        required=True,
         help="Path to the text dataset file, Type string, File type output",
-    ),
-) -> None:
-    """extract_pdbids_drugbank_xsdata."""
-    logger.info(f"drugbank_xml_file_path: {drugbank_xml_file_path}")
-    logger.info(f"smiles: {smiles}")
-    logger.info(f"inchi: {inchi}")
-    logger.info(f"inchi_keys: {inchi_keys}")
-    logger.info(f"output_txt_path: {output_txt_path}")
-
-    extract_pdbids_drugbank_xsdata(
-        drugbank_xml_file_path=drugbank_xml_file_path,
-        smiles=smiles,
-        inchi=inchi,
-        inchi_keys=inchi_keys,
-        output_txt_path=output_txt_path,
     )
 
-
-if __name__ == "__main__":
-    app()
+    args = parser.parse_args()
+    main(args)

utils/extract-pdbids-drugbank-xsdata-plugin/tests/test_extract_pdbids_drugbank_xsdata.py

@@ -4,6 +4,8 @@
 from polus.mm.utils.extract_pdbids_drugbank_xsdata.extract_pdbids_drugbank_xsdata import (  # noqa: E501
     extract_pdbids_drugbank_xsdata,
 )
+from sophios.api.pythonapi import Step
+from sophios.api.pythonapi import Workflow
 
 
 def test_extract_pdbids_drugbank_xsdata() -> None:
@@ -15,6 +17,37 @@ def test_extract_pdbids_drugbank_xsdata() -> None:
 
     input_xml_path = str(Path(__file__).resolve().parent / Path(input_xml_path))
 
-    extract_pdbids_drugbank_xsdata(input_xml_path, [], inchi, [], "out.txt")
+    extract_pdbids_drugbank_xsdata(input_xml_path, [], inchi, [], "output.txt")
 
-    assert Path("out.txt").exists()
+    assert Path("output.txt").exists()
+
+
+def test_extract_pdbids_drugbank_xsdata_cwl() -> None:
+    """Test extract_pdbids_drugbank_xsdata CWL."""
+    cwl_file = Path("extract_pdbids_drugbank_xsdata_0@[email protected]")
+
+    # Create the step for the CWL file
+    extract_pdbids_drugbank_xsdata_step = Step(clt_path=cwl_file)
+
+    input_xml_path = "drugbank_10_fake_records_5.1.10.xml"
+    input_xml_path = str(Path(__file__).resolve().parent / Path(input_xml_path))
+
+    inchi = ["InChI3491", "InChI8564", "InChI7556"]
+
+    extract_pdbids_drugbank_xsdata_step.drugbank_xml_file_path = input_xml_path
+    extract_pdbids_drugbank_xsdata_step.inchi = inchi
+    extract_pdbids_drugbank_xsdata_step.output_txt_path = "output_cwl.txt"
+
+    # Define the workflow with the step
+    steps = [extract_pdbids_drugbank_xsdata_step]
+    filename = "extract_pdbids_drugbank_xsdata"
+    workflow = Workflow(steps, filename)
+
+    # Run the workflow
+    workflow.run()
+
+    # Check for the existence of the output file
+    outdir = Path("outdir")
+    assert any(
+        file.name == "output_cwl.txt" for file in outdir.rglob("*")
+    ), "The file output_cwl.txt was not found."